How to Get Ahead of 99% of Data Scientists with Streamlit (Tips from Tyler Richards)
Summary
TLDRIn this Data Science podcast episode, Tyler Richards shares his journey from election science to becoming a data scientist at Snowflake. He discusses his experience at Meta, focusing on content integrity and the challenges of moderating vast amounts of user-generated content. Richards also delves into his passion for Streamlit, an open-source library that simplifies the creation of data apps. His enthusiasm led him to write a book on Streamlit, which he continually updates to reflect the evolving tool. The conversation highlights how Streamlit can enhance job applications and the potential for data scientists to monetize their work, with Richards' recent creation of the 'stpaywall' component, aiming to facilitate online monetization for Streamlit app creators.
Takeaways
- 😀 Tyler Richards, a data scientist at Snowflake, co-hosts the podcast and discusses his journey from election science to working at Meta on content integrity, and eventually joining Streamlit.
- 📚 Tyler authored a book about Streamlit, which he revised and updated after joining the company, aiming to create the resource he wished existed when he started learning Streamlit.
- 💡 Streamlit is positioned as an effective tool for data scientists to stand out in the job application process by creating shareable, interactive apps that demonstrate their skills and projects.
- 🔑 Tyler's experience at Meta involved working on complex machine learning models for content moderation, emphasizing the importance of balancing free speech with platform integrity.
- ✍️ Writing a book was a method for Tyler to consolidate his learning and create a comprehensive guide for others to learn Streamlit effectively, focusing on deep understanding rather than fragmented knowledge.
- 🤝 The value of mentorship and feedback is highlighted as crucial for Tyler's growth in programming and data science, suggesting that pair programming and code reviews are beneficial practices.
- 💼 Tyler's transition from Meta to Streamlit was driven by a desire to be part of an early-stage company and to work more closely with a product he was passionate about.
- 🌐 Streamlit's open-source nature allows it to be used across various platforms, not just within Snowflake, providing flexibility for developers to host their apps in various environments.
- 🛠️ Tyler introduced 'streamlit_paywall', a component to facilitate monetization of Streamlit apps, with the goal of enabling data scientists to generate income from their work online.
- 💡 The importance of curiosity and kindness is emphasized as core values that Tyler would advise his younger self to embody, suggesting these traits are fundamental to personal and professional development.
Q & A
What is Tyler Richards' background and how did he become a data scientist?
-Tyler Richards has a background in election science, having worked in nonprofits and academia. He transitioned to data science and worked at Meta (formerly Facebook) focusing on integrity-related projects. His interest in data science was further fueled by his experience with the open-source library Streamlit, which he used extensively and eventually wrote a book about.
How did Tyler Richards and the host end up working at Streamlit?
-Both Tyler and the host were initially internet friends who connected through their shared interest in data science. They both joined Streamlit after recognizing its potential and enjoying their experiences with the platform, transitioning from being just friends to co-workers.
What was Tyler Richards' role at Meta focused on?
-At Meta, Tyler worked in the Integrity space, focusing on content moderation. His role involved developing machine learning models to efficiently identify and remove content that violated Meta's rules, balancing user experience, business interests, and free speech considerations.
What motivated Tyler Richards to write a book about Streamlit?
-Tyler was inspired to write a book about Streamlit after publishers reached out to him due to his online writings about the library. As an avid reader, he saw it as an opportunity to create a resource that he would have wanted to read, focusing on practical applications and experiences with Streamlit.
How did Tyler Richards' experience at Meta influence his decision to join Streamlit?
-Tyler's experience at Meta, where he worked on large-scale data science projects, made him appreciate the efficiency and capabilities of Streamlit. His desire to work in a more unpredictable and fast-paced environment, as well as his love for the product, led him to join Streamlit.
What is the significance of Streamlit in job applications according to Tyler Richards?
-Tyler believes that creating Streamlit apps can significantly enhance a job application by demonstrating practical skills and creativity. By building an app tailored to a specific job or company, applicants can stand out and showcase their ability to create tangible, impressive projects quickly.
How did Tyler Richards learn Python and what was his initial programming language?
-Tyler initially learned R during his stats classes and used it extensively. However, when he was offered a data science internship at Procter and Gamble, he had to quickly learn Python. He scrambled to learn Python in about three weeks and further honed his skills through pair programming with experienced engineers.
What advice does Tyler Richards have for finding a mentor?
-Tyler suggests that mentorship should be a two-way street where both parties benefit. He emphasizes the importance of showing earnestness, intellectual curiosity, and the potential to reciprocate the help in the future. He also highlights the value of hands-on learning and feedback from those more experienced.
What is the purpose of the 'stpwall' component Tyler Richards created?
-The 'stpwall' component is designed to simplify the process of monetizing Streamlit apps. It allows developers to easily implement a paywall in their apps, enabling them to charge users for access. Tyler's goal is to encourage more data scientists to create and sell their apps, potentially making a living from their work.
What advice would Tyler Richards give to his younger self?
-Tyler would advise his younger self to be more curious and more kind. He believes that embracing curiosity and treating others with kindness are essential qualities that have helped him in his personal and professional life.
How can listeners reach Tyler Richards and learn more about his work?
-Listeners can reach Tyler Richards through Twitter, where he is most active. His Twitter handle is Tyler J Richards, and his email can be found on his Twitter profile. He also mentions that the proceeds from his book are donated to charity, indicating his commitment to giving back.
Outlines
📈 Transition to Co-worker and Data Science Journey
The host welcomes Tyler Richards back to the data science podcast. Tyler introduces himself as a data scientist at Snowflake with a background in election science. He discusses his transition from internet friends to co-workers with the host after both joined Streamlit. Tyler shares his experience at Meta, focusing on content integrity and the challenges of content moderation at scale. He also talks about his journey of writing a book on Streamlit, which led to his current role as a data scientist at Streamlit.
🤖 Content Integrity at Meta and Machine Learning Models
Tyler delves deeper into his role at Meta, where he worked on content integrity, ensuring harmful content is removed from platforms like Facebook and Instagram. He explains the necessity of creating efficient machine learning models to automate content moderation due to the vast volume of uploads. Tyler also touches on the complexity of defining and moderating hate speech and the impact of these decisions on user experience and business.
🚀 The Appeal of Early-Stage Companies and Joining Streamlit
Tyler discusses his desire to join an early-stage company for the opportunity to work in a fast-paced and unpredictable environment. He was particularly drawn to Streamlit due to his love for the product and the chance to contribute to its growth. After writing a book on Streamlit and maintaining a relationship with the founders, he was invited to join the company as a data scientist, which he gladly accepted.
🔍 Leveraging Streamlit for Job Applications
The conversation shifts to how Streamlit can help job seekers stand out in the application process. Tyler shares his strategy of creating Streamlit apps tailored to specific job applications to impress recruiters. He suggests that creating an app for a potential employer demonstrates coding skills and initiative, setting candidates apart from the competition.
📚 Writing a Book on Streamlit and the Editing Process
Tyler describes his process of writing a book on Streamlit, emphasizing the importance of dedicated time and focus. He explains that the second edition of his book incorporated lessons learned from daily use of Streamlit, aiming to help readers avoid common mistakes. Tyler also discusses his learning style, preferring in-depth focus over short bursts of information.
🛠️ Streamlit for Work and Overcoming Deployment Challenges
The discussion moves to using Streamlit in a professional context, particularly after Streamlit's acquisition by Snowflake. Tyler outlines the benefits of Streamlit for Snowflake, which simplifies deployment, security, and data sharing. He also acknowledges that Streamlit remains an open-source project, free to use and deploy on various platforms.
👨🏫 Learning Python and the Value of Mentorship
Tyler shares his personal journey of learning Python, initially starting with R and then quickly adapting to Python for an internship. He highlights the importance of mentorship and receiving feedback from more experienced programmers, which significantly improved his coding skills.
💡 The Concept of STP Wall and Monetizing Data Science Projects
Tyler introduces STP Wall, a component he created to simplify the process of monetizing Streamlit apps. He expresses his belief in the potential for data scientists to generate income from their projects and hopes that STP Wall will encourage more developers to do so by making it easier to accept online payments.
💌 Advice to Younger Self: Curiosity and Kindness
In the final segment, Tyler reflects on the qualities he values most: curiosity and kindness. He advises his younger self to cultivate these traits and shares his belief that embracing curiosity through Streamlit and treating others with kindness are key to personal and professional growth.
Mindmap
Keywords
💡Data Science
💡Streamlit
💡Election Science
💡Meta (Facebook)
💡Content Moderation
💡Machine Learning
💡Snowflake
💡Personal Projects
💡Monetization
💡GitHub
💡Curiosity and Kindness
Highlights
Tyler Richards shares his journey from election science to data science at Meta and Snowflake, highlighting the transition from academia to industry.
He discusses the appeal of working with Streamlit, an open-source Python library for building custom web apps, and how it influenced his career move to Streamlit as a data scientist.
Tyler's experience in election science and content moderation at Meta provided a strong foundation for his work in data integrity.
The creation of Streamlit for Snowflake is detailed, emphasizing its ease of use for data scientists to share and deploy apps within a secure environment.
Anecdotes from Tyler's book writing process reveal insights into learning and teaching data science through Streamlit, and the value of focused study sessions.
Tyler's approach to job applications using Streamlit apps to stand out, showcasing his unique method of impressing recruiters with interactive projects.
He explains the importance of mentorship and receiving constructive feedback to improve programming skills, drawing from his experiences with pair programming.
Tyler's perspective on using Streamlit for work, beyond just deployment, includes its role in data analysis and the benefits of open-source accessibility.
The development of the 'sc-paywall' component by Tyler is explored, aiming to simplify online monetization for data scientists and content creators.
His vision for Streamlit as a platform for data scientists to create and sell apps, similar to the App Store model, is shared.
Tyler's advice for finding mentors involves creating win-win situations and showing intellectual curiosity and earnestness.
The importance of curiosity and kindness as core values in personal and professional development is emphasized.
His experience learning Python in a short time for a data science internship and the role of determination in skill acquisition.
Tyler's reflections on the power of feedback loops and the impact of working closely with more experienced programmers.
The role of Streamlit in democratizing data science by allowing anyone to create and share data apps without extensive coding knowledge.
Tyler's final advice to his younger self and the audience, focusing on the importance of continuous learning and maintaining a beginner's mindset.
Transcripts
so welcome back to another episode of
the data science podcast and today I
have the fortunate pleasure of having
Tyler Richards back on the podcast Tyler
would you like to introduce yourself
yeah yeah hey uh so good to be back um
yeah it's been it's been a hot minute
and I really enjoyed the the last time
that I was on here and then um basically
after after uh we recorded last time
both of us ended up joining streamlet so
now we're co-workers so before we were
just internet friends and now we're
co-workers and what a fun what a fun
transition that is so as you mentioned
I'm Tyler um I am a data scientist that
currently works at um at snowflake um my
background is in election science and I
did election science at nonprofits and
um at uh while I was at University a
little bit afterwards as well and then I
went over to meta to continue on uh
Integrity related work um as I um worked
more at meta I um came across open
source library streamlet and honestly
just really liked it I tried to bring it
inside of meta played around with it
with person personal projects and um
really just got enamored with the
library as a whole um as I was doing
that I started writing um online more
about my experience with uh streamlet
and then I uh a couple book publishers
got in contact with me and were like hey
uh your writing's a little bit about um
streamlet do you just want to run like
write a book and I was like okay that
that sounds great I'm a I'm a big reader
a pretty voracious reader and um so I
thought it was like yeah this is kind of
the I wanted to write the book that I
always um wanted to read and so I did I
did that I I published a first version
and then I went from after I published
the first version I joined stream as a
data scientist um and then I uh just
recently edited um and uh released a
second version uh of the book uh a few
weeks ago so that's the um the the gist
of it um uh yeah yeah that that that's
pretty much me that's a long
introduction but that's pretty much me
yeah awesome and and yeah definitely
check out the first episode uh of our
part podcast um if you haven't already
and yeah so I I guess you have already
provided um like your the the background
of who you are and then um your your
prior row over at me meta um could could
you maybe um go into a little bit depth
into like um what what did you do at
meta and then
and the thought process of actually
joining stret yeah So Meta I actually um
I have um nothing but positive things to
say about about my time at meta I know
that might be a little rare but I also
left right after um or right before uh
all of the kind of cycle that they've
been going through with um uh
optimization and cost optimization and
layoffs and um the stock price going
down and all this other like there's
definitely been a big Vibe shift in meta
um after I left but um I left long
before a lot of that was happening and
so I had a relatively um I had a really
good time I learned a lot about how to
work within a big organization a lot
about how to do data science how to work
with engineering partners and product
partners and um really learned a whole
lot there uh my actual role was in the
Integrity space and so lots of stuff
gets uploaded to Facebook and Instagram
um every single day um every single
second and and there is a very very
strong desire to get the worst of the
worst content off of um off of these
platforms um both because it's a bad
user experience and um it's also bad for
business as well like no Advertiser
wants to have their ad showing up right
next to hate speech or right next to
nudity or right next to um like
terrorist content or all this other all
these other related things um and so
there's huge teams at meta that are
frankly the best in the world at taking
um content down um that breaks meta's
rules and because there's so much
content you can't just pay people to to
um take down the content to like look at
it and take it down you have to create
these like very very efficient stable
machine learning models that will look
at a piece of content and take it down
um and then I'll also be able to have to
be able to say Hey you know we're 85%
sure that this is hate speech that this
post is hate speech why don't we go and
we send it to uh an expert that we know
is really good at detecting um if this
is hate speech or not and you know the
next piece of content might be oh we're
actually 92% sure that this is hate
speech we might actually take that one
down because we think that that it's
going to get 10,000 views in the next
you know in the next 30 minutes or
something so those are the sorts of like
very complicated questions because
they're they really were the
intersection of like social science and
the questions of well what does it mean
to have a rule um on Facebook like what
does it mean for us to preserve free
expression and Free Speech what does it
mean for the the business community on
Facebook what does it mean for different
populations um like what even what is
hate speech what is hate speech is a
really like hard question um and so my
main projects were focused on
measurement so are we doing a good job
at that at all that work or are we doing
a bad job um that like is really what it
came down to um that work was man I mean
the the pros and cons of that work are
are are definitely sharp I would say
everyone that I talked to while I was a
data scientist there had an opinion
about meta Integrity or this is when it
was called Facebook so this is Prem meta
but everyone had an opinion everyone was
like oh you guys are doing a blank job
at this you know either great or
terrible or they' be like oh you take
down too much content you know and then
I I would be in California right I'd be
in California people be like oh my God
you guys are not taking down nearly
enough content um you need to be taking
down blank blank blank whatever whatever
side that I don't agree with you know
you need to take down all their stuff
and then I would go back to and hang out
with my family in Florida and I would
you know see a bunch of you know friends
that I grew up with uh and then they
would be they would all say oh you guys
are taking down way too much content in
you know X Y and Z area and so there was
there no one was ever neutral about uh
how we were handling uh how we were
handling speech at meta so that was fun
because everyone has an opinion on your
job um and then annoying because I have
the same conversations about my job with
everyone uh and yeah so that that was my
my background at meta it was really
useful to be in election science
beforehand because I already had a lot
of background in um handling those
problems at a much smaller scale but I
mean with meta you just get like you're
like oh if I change the Precision on
this machine learning model by .1% that
literally over the course of the Year
might affect billions of posts um so
yeah wild wild place if you're looking
for scale there's no place like meta
I'll tell you what I mean Google Google
but but Google and meta are like
scale yeah sounds pretty amazing like
you know like the the work that you were
doing and yeah
hard the entire population of the world
it really does it's like basically
almost everyone I mean right now there
are more smartphones than there are
people on the face of you and and more
than half of people in the world use
Facebook every month I think that's
probably about right if I'm not exactly
right I'm close um or use a Facebook
product if you combine WhatsApp Facebook
and Instagram you're like you're
touching almost everyone outside of
mainland China that's pretty much it
it's wild yeah
definely and yeah could you walk us
through through your thought process of
you joining streamlet yeah yeah so as I
was at meta and kind of one of my one of
my conent there's like two sides to this
right where side number one is I had a
strong desire to join a very early stage
company because within me within meta I
had like I definitely felt free to do um
I felt free to you know work on
Integrity related problems and integrity
measurement related problems but if
there were like other problems that I
thought were really pressing I kind of
had to stay in my Lane which is good
like the whole point of being uh of a
very good large tech company and large
business is that you produce
organizations that are predictable like
I can predict what this organization is
going to produce in three to six months
because otherwise just makes it very
hard to run the business as a whole and
the point of a very small startup is to
do some things that are a little bit
unpredictable so I you want to you know
have these big step changes in
production you want to be able to move
people at a moment's notice into a more
critical area of the company um and so I
had to I I just wanted to be uh in a
space that was just a little bit faster
um and I wanted that that experience for
myself um in terms of streamlet
itself I mean I I join because I love
the product like I used the product and
I said this is awesome I don't want to
be creating data science work outside of
streamlet ever again so the number one
way to guarantee that no one will make
me use a tool that is other than
streamlet is to work at streamlet of
course I'm just gonna be creating stream
tools all the time um yeah so I uh I I
join because I I I love the product um
when after I I wrote the book um you
know some of the the founders reach out
I I would you know chat with Adrian you
know once a month or we had like a
bi-weekly call where we talked about
feedback and you know all all this other
um stuff and yeah so then uh when they
started hiring data scientists uh they
were like would you like to join and
yeah yeah yeah I would uh so that that
was the I wish it was like more
difficult or like a oh there was it was
super complicated and whatever else but
I mean I love the product I believed in
it a lot and this company was out there
wanting
to increase the number of streamlets
that were out there so it was like very
much a perfect fit for for the thing
that I was looking for so I'm just I'm
glad they said yes I'm glad glad you
know otherwise I would have just been
like all right well I guess I just like
stream from afar which would have been
okay but it was very good fit for me
awesome yeah uh like I I do remember um
uh I guess we join Str about
approximately the same time right yeah
yeah I guess I left Academia in November
and then end of November I joined
TriMet 2021 yeah yep yep exactly it was
like
very yeah I mean you're in exactly the
same boat right where you were like hey
I really like this product exactly I
want more people to be using this I want
to use this on a regular basis like it
totally Mak
sense same thing with a few other people
right like johannas is a great example
where he made a really cool app um the
the GitHub Stars app was excellent um
and that the machine learning generator
app yeah exactly machine learning
generator app that thing's amazing that
could have been a small business all on
its own um like so cool so for for
everyone listening he had these two apps
one of which um he finished in like a
few days over like break where it just
like looked through all of your GitHub
history and then it like ran a nice anal
is on your GitHub history um and then
showed it to you and then you could
tweet about it and that went that thing
went so viral like so fast I saw it
everywhere all over Twitter people were
like looking at you know entering their
own GitHub username and just going after
it it was awesome and he made this other
app that is like this machine learning
generator where you could kind of upload
a data set or upload um you know click a
bunch of different like questions and
then it would create all of this python
code that would train and then use and
then infer um a machine learning model
for you based on your data set so I was
like this is aw this is so sick yeah
definitely and yeah I mean we're we're
talking about like how we join strem Le
and then uh how we love strem um yeah so
I guess this brings us to the or our
next question is which is
um H in your opinion how can stre help
one find a job role yeah the easiest the
the way that I think about it now is
basically streamlet is an unbelievable
way to stand out in a job application
process or before a job application
process um so as I was like you know
working at meta every every you know six
months or something I would go out into
the job market and just kind of like see
what was out there and you know test the
waters and I found that and I don't know
if other people like have this same
feeling but I always felt like just
applying to a th000 jobs with a wor a
blanket
resume just like didn't work like my hit
rate is just low and I don't want to get
like just one of the Thousand jobs that
I'm applying for I I want a a set of
jobs that there's like five jobs that I
really care about and I would really
love to work at and so for me it made so
much more sense to spend more time on a
smaller number of applic that I cared
about rather than like almost no time on
a bunch of applications that I like most
of which I couldn't even tell you that
much about the company so um when I was
doing that I was like all right well how
do I stand out in front of a recruiter
okay I can have a cool GitHub but
recruiters can't code so if I show them
like oh there's this awesome python file
or Jupiter notebook there's no way
they're going to know if it's if it's
impressive or not right like they're
going to make three or four clicks get
into a jupter notebook and then wow they
used Panda really efficient like there's
no way they're going to be able to
assess if I'm good or not so then I'm
like all right well I went to a I went
to a US University which was definitely
a benefit but the University of Florida
where I went to wasn't it's not like a
you know one of the fancy private
schools it's not like a Harvard or a
Yale or something so I don't have that
brand name that would make everyone say
all right well we got to interview this
guy now because he went to Stanford um
so I didn't have that I'm like all right
well how do I like what and then I I
worked at Mena but even before I worked
at meta it was like I had worked at
Proctor and Gamble which is cool but no
one thinks of them as like wow they
produced the best data scientists in the
whole world like no one no one thinks
that so I'm like all right well how do I
stand out and I realize well if I can
create a data product that someone can
click on and see something cool really
fast
then then they'll be impressed and
they'll be like wow they created this
for me how did they create this whole
website just for me like it's an app and
I can mess around with it like that's
awesome um and with streamlet it's
honestly not that hard right where you
can create five apps that you know
whether it's you know some data science
project that you've done already where
it has some nice markdown some nice
explanations a cool map something that
people can click on like that is doable
for you know a small set of applications
um where if I was like before stream if
I was trying to show off something and
create a whole website that would be or
like you know oh I create an awesome API
or like ajango app or something it would
be like that's too much work for an
individual application so like the way
that I saw and my hit rate of like
applying for jobs
or once I was in the job application
process usually there's like a data
science project part of it just doing
that in streamlet and then sending them
the link um that that hit rate was
really high people were very impressed
by it um and so that's kind of just like
a little bit of alpha in a way that
hopefully so many people start doing
this that in the long term like it's not
uh you know every everyone expects to
people to create streamlets as part of
their like uh ja application process but
for now I think there's a lot of Alpha
and just
hey I'm going to create an app for you
and then like just imagine it from a
recruiter's perspective I I don't have
to see a lot of their code I know that
they can create something cool because
created something cool and here it is
like it's right here this person
definitely knows how to code um yeah so
that's that's the main like way that I
think about it at the moment so strongly
encourage people to just just try it out
like just try it with two two companies
right where find a way to make stre app
for them someone that you really want to
work for find someone of the company
that you think you're going to click on
your link email them or find them on
LinkedIn or something and say hey I saw
you had this job opening for United
Airlines data analyst um I pulled the uh
like flight history of the past thousand
flights and you know found this cool
model that predicted blank here it is or
something whatever it is I just did a
cool visualization that I um that I you
know that I thought was pretty and help
help me understand you know flight
patterns in the United States better or
something like whatever it is just
something like that is g to make you
like it's gonna make you so much better
than anyone else that's applying to
these jobs most people that are applying
to these jobs put no effort in they just
toss a resume in a pool and they leave
it so like you can be in the top 1% of
jobs very easily um of applications very
easy so I don't know I don't know if um
well I mean like I obviously haven't
gone around searching for jobs in in a
hot minute but um uh but yeah that
that's like how I started to do it and
and one thing that I talk about I have a
chapter in the book about like hey
here's an actual uh thing that I did
that helped me get an offer at a place
um yeah so that that's the uh the gist
bit yeah I mean that that's like a hack
that uh I think not a lot of people know
about it and you know like the listeners
now you know and yeah it's like it's
like it makes me think of kind of like a
cheat code you know like how you could
use that to leverage and you know up up
your profile and provide instant value
to the to the company that you're um
interested in applying for yeah I mean
that that's certainly a great tip
um yeah you you touched
upon writing your book um could you go
over the process of writing a book and
the thought process of you actually
writing the book and like the daily
routine and then finally maybe would you
consider writing another book yeah you
already have two books yeah so I would
say the process the process was actually
remarkably straightforward where you uh
at least for me what made sense for me
is um I designate time in my day and I
say okay the only thing I'm going to do
for this hour or this two hour block or
this 45 minute block is blank and then I
worked until that time was over and then
I left and then I did that I didn't just
like you know do this forever and so I
just had designated times throughout the
week where I'd be like all right this is
my writing time this is what I'm going
to do um the process of editing the book
um like for the for the second version a
lot of that was stuff that I'd been you
know thinking about and so to give you a
bit more background when I wrote the
first version I used streamlet as as
kind of a hobby where hey this isn't
this isn't something that I'm doing
every day all day but this is whenever
you know whatever I can whenever I can
find an excuse to make a streamlet app
I'm going to make it and so my number of
hours making streamlet apps was higher
than most people for sure um but then
after I joined streamlet
I
spend all of my time at stream is a
primary data product that we produce
like that's the only thing that I
produce these days it's like sometimes
I'll make a notebook but most of the
time I'm going to make a stream lip and
so for the second version my goal was
okay can I take all the things that I've
learned from making streamlets every
single day for the past year and a half
and can I like add that to the book in a
way that beginner and an intermediate
person can understand easily so that all
the mistakes that I've made they don't
have to make they can they can they just
they just yeah they don't have to make
these mistakes I can just figure it out
um yeah so that was the kind of like the
process of learning enough to be able to
have stuff to write about um and then
the actual writing experience is just
time so you just as long as you're
willing to dedicate time you could do it
um I've kind of
like I think that the reason I wrote it
is kind of what I said before which is I
I believe for me the best way for me to
learn is to sit down and focus on one
thing for an extended period of time I
do not learn very well from 20 minutes
of this don't touch it for two days 20
minutes of this don't touch it for six
days 20 minutes of this don't touch it
and and I don't I just don't really
learn well that way I know a lot of
other people that do and a lot of other
people learn well from audio or from
video um like think of the number of
people that you've helped with this
YouTube channel right where so many
people learn from video tutorials and
they love it and that helps them a whole
lot um I don't that wellest say um so I
I do a bit but I don't really that I
don't that much right so for me I was
like all right well I want my ideal
learning setup is long form it's focused
and it takes you a to z right that's how
I want to learn um and for streamlet
there was kind of no way of doing that
um and so that was the main reason and
so the reason for version two is really
stream was on
0.87 and um like the the book the first
version used 0.87 or 0.67 I for exactly
and now streamlet is on
1.26 1.27 maybe
1.28 um so there's been 50 updates since
since in that in that time um and so
yeah there's so many different features
that are just released and I have all
this new experience from using stream
life uh and learning about and and just
like using it every day so that's that's
the main like my main thought for for
writing the second version the question
of like whether or not to do it again is
really hard um for me I
think I'm very excited for lots of other
people to make really awesome streamlet
books and as lots of other people are
making really awesome streamlet books um
then I'm I'm excited to uh support those
stream books and not have to write like
more stream books uh I especially as a
data scientist I'm like okay well this
is this is great think like I enjoy
doing data science a little bit more
than I do writing um and so I I kind of
doubt that that I will like continue too
long in in the writing sphere um so I'm
not I'm not exactly sure but my my goal
is that I get to support lots of other
people writing really awesome streamlet
books and tell everyone go you know go
and look at those instead of mine you
know two years from now or
something cool yeah I mean certainly
your your book is amazing and it's a
great resource for anyone you know
looking to get started and um
yeah yeah I mean I was very fortunate to
be able to look at your book and read
the entirety of it and I do say it's
it's kind of like having a mentor you
know like sitting beside you because
like the the way that you you've written
it is it's not like a typical book it's
more like you're talking to a friend and
then you're guiding your friend like
step by step and then you're
incrementally U building up the app and
then you're you're you're taking us
along the thought process of here here
is a new feature and then why are we
adding it here and what's like the
caveat of that and how we could
incrementally um improve upon it yeah I
wonder when we were going to get to that
point when uh we revealed that you are
the technical editor of this book which
is so kind and so nice and you spent so
much time on it and did such a good job
like there are so many typos and edits
and like like like code stylistic things
and like the fundamentals of it are so
much better because of the work that you
did and so it's like I mean I just like
you know I try to thank you every every
time I can but I'll thank you to the
ends of the Earth I really appreciate it
it's so nice yeah you certainly my
pleasure and yeah I I get a new chance
to learn more about streamlet and you
know it's kind of interesting to you
know like when you already know about
something but then when when you're
relearning it um you get to see it at a
different angle and in doing that it it
kind of you know provides fresh
viewpoints to the existing technology
that you are already familiar with but
then not really familiar in some aspect
and yeah I guess it's it's a great way
to to learn in a holistic way yeah the
way that I've I've explained it in the
past of like my learning style and the
way that I try to like the way the way
that I try to approach this book is that
that there's a lot of like a lot of
people really like kind of a Swiss
cheese method of Lear where you have a
bunch of different slices of cheese and
there's holes in each one but then if
you lay them all on top of each other if
you lay enough slices you're going to
double learn you know there's going to
be sections where there's every section
has no holes like if you just like look
straight down on it because you know you
get a little bit from each one and a lot
of people love to learn that way I do
not I just want one slice of cheese that
has no holes in it so so the uh that
that's that's what I'm most interested
in so I tried to like write a book that
just gives you one slice of cheese no
holes not really the Swiss cheese method
um but I don't know if that makes any
sense as an analogy or not but that's
like how I try um I've tried to think
about it yeah love love the analogy um I
mean when when we're we're talking about
this it kind of lead me to think about
like um what are your advice on how one
could use TriMet for work
yeah yeah well so there's there's a
couple different things right so in s
since version one uh stream got acquired
by snowflake um and so that's why uh
both of us currently work for uh for
snowflake you've got a snowflake shirt
on perfect um yeah exactly perfect uh so
that we have this product inside
snowflake that we just called streamlet
for snowflake um and the what that does
is it essentially will handle all of the
the difficulties in hosting as well as
data connection and sharing and you know
all the annoying things um about like
deploying apps within work are mostly
related to can I get my it team to set
up an ec2 instance and then host it and
how do I do all the security around it
and how do I get access to my data and
make sure my data has access to this how
do I make sure I'm not actually
accidentally leaving some Port open and
then someone else can access it and I
didn't even know
um like all of that stuff is is kind of
difficult um and so snowflake just said
hey we'll figure it out for you like
we'll do all that annoying stuff so you
don't have to touch it um and you can
just go into your snowflake account and
start writing python code and there it's
your streamlit and you can press a
little share button and then great you
can share it with your colleagues your
product managers whoever else um and so
honestly that's the hardest part
um about like using streamlet for work
is all of the work rated security
hosting
things um that said uh that's probably
like the biggest the biggest pitch um
I'm also biased pitch person because uh
that's the product that I work on right
like that's that's what I've spent the
past like you know nine months uh
messing around with to try and get this
product state where like we're happy
with it and we think it's like the best
way to use streamlet at work that said
stream is an open source project stream
is going to remain an open source
project it is free to use you can grab
it you can go and host it on your own um
on your own instance you can host it on
a hugging face space hugging face has
hugging face for work too so you can
make streamlet apps inside of hugging
face you can do it inside of Heroku you
can do it right inside of AWS right if
you want to have all the knobs and be
able to you know make all of the little
intricate decisions yourself you
absolutely can do that and I think uh
there are lots of places online where um
you can uh like learn how to deploy an
app on AWS there's a section in my book
about how to deploy an app on Heroku um
and also for for hugging face as well so
the uh I'm not I'm not here to say oh
the only way that you can like use
stream letter it work is
um is is in is inside of snowflake like
no that's that's that's not what it is
and it's not the intent um uh the intent
either um so yeah I um
I would say that's the biggest uh the
biggest way to use TriMet for work is
like the the hardest thing is figuring
out deployment security sharing um you
know where does my data sit all of that
stuff right yes certainly and yeah so
yeah Str is like ubiquitous and it's not
only on on on you know like a specific
platform but everywhere
and yeah
um yeah so I guess it's now time to move
on to the next question uh which is if
you were going to learn stret again how
would you do it yeah well for I mean for
it it depends on it depends on
how
um let me think the the honestly the the
the the actual question the actual
answer to this is the first place I
would go is I would go to docs.
stream.io mhm at that's our where our
docs page is our docs page is amazing
it's really good um there's a bunch of
people there that do an excellent job
about creating and maintaining just like
a delightful documentation page um and I
would start there before I would even
think about going to all the different
ways um to to to consume streamlet
content and a learn streamlet so I would
pip install streamlet I would look at
the docs and I would try it out myself
and just see how it feels
um and I would you know go through the
different widgets and um maybe there's a
little getting started section in the
dogs that I think is really good that
just like oh here's how you pip install
something here's how you you know start
a you know create a new python file
here's how you run it as a streamlet app
like all that really basic stuff um and
so I would start there and then if I
felt oh I'm actually missing a lot of
the ecosystem I actually you know I I if
I really like to read and and read
reading was my best way that I wanted to
learn I would go and I I would you know
buy my book and and read it that way
right that that would be the the way
that I would go if I um if I really
liked uh like consuming video content um
I would go to to your channel and then I
would also go to Fino's Channel Fino's
channel is really good too um I think
both of your channels are like I've
definitely watched quite a few of the
videos and learned something new and I
think it's a great um a great and
entertaining way to learn as well um so
if I was doing video content I would
definitely go to those two places I've
seen some other YouTube channels I
forget exactly we can like link them
below as well but I think avra has
YouTube channel that is like quite good
um I I enjoy that um I've seen a couple
videos there and then I know there are
some other um some other videos that I'm
kind of forgetting right now but I like
I really like all of them um so I'm I'm
a fan so I'd start there and then my
main thing is I would find find a
project that I cared about and then just
tried to make it at streamless so
whatever it is just have a little
project and it like learning makes
learning is more fun when you're trying
to
learn to do something if you're just
trying to learn in a vacuum you're like
well I'm just learning to learn it's
kind of like well like why am I doing
this like I if I have something in front
of me that I want to create an app to
blank I want to analyze my good reads
history I want to I went to create an
app so that other people can analyze
their good reads history that was my
first project in stream life you know uh
what we were talking about with yanas
like he was like oh well I want to see
my GitHub history great I want to create
an app so other people can see their
GitHub history awesome those are perfect
that's exactly what you should be uh
where you should really be looking for
like find a project mess around with it
see if you can do it and uh do something
manageable and have fun with it that's
my that's about be my go-to yep and and
a pointy not is that the the good read
app that you've created I think is it
went viral and it's amongst one of the
top uh Pages uh for strs yeah yeah I
would say it's a probably a little
biased because it's on our
website exactly IO and so if uh so if
people click on that so I think that is
like a big source of the virality but
yes it it it went uh it went kind of
viral when I released it at first um
yeah that went that went well right yeah
think that people uh read these days so
a lot of people to analyze their
Goodreads history yeah or we analyze
their YouTube viewing history right yeah
or sptify oh yeah uh Tyler Simons
another data scientist that that works
on our team made a very nice Spotify um
app that I think is really cool we can
um link it in the comments as well um
but it's pretty it's it's a really
beautiful app all amazing yeah it's it's
kind of like when whenever we have IDI
wild ideas and then it's like you know
we could make a s that yeah yeah why not
I mean it's python right so like python
it's like what what can you do in Python
um pretty much anything pretty much
anything like as as long as it's you can
do it on a computer like it's turn
complate language like you're fine like
you could probably figure it
out yeah yeah speaking of which um how
did you learn python did you learn it in
school or did you so taught oh man I
learned python well all right so the the
actual answer is I started with r um and
I learned R and I did that because I was
in a lot of stats classes and those
stats classes use R um and so that was
like my main language my bread and
butter and then I went to Proctor and
Gamble and as I got to Proctor and
Gamble basically I was like uh I had
interviewed with them and they had
accepted me and I was originally going
to be a data analyst and I wanted to be
a data scientist but I was going to be a
data analyst and then um got this phone
call like a few weeks before my
internship they called me and it was
this guy named John his name is John
rool he was one of the data scientist
there and he called me and he was like
hey Tyler I'm gonna be your intern
manager um uh but you know we're trying
to figure out if uh we should have you
as a data science intern or we should
have you as a data analyst intern and I
was like okay uh and he like do you want
to be a data science intern or data
analyst insurance and I was like well I
want to be a data scientist so I think I
want to be a data science insurance he
was like okay do you know python on and
I was like uh not really and he was like
oh well then I don't know if that's
going to work and I was like oh but I
can learn I can learn I can figure it
out and it was like I was like I know R
like RS will be close and he was like
okay well if you can figure out python
then you can be a data science enter I
was like okay I'll figure out python
that was it and then I was like oh my
God I gotta figure out Python and then I
just scrambled around trying to Learn
Python for like three weeks um and then
so I just you know went online read all
the books that I could figure out like
kind of just like freaked out um and
scrambled and scrambled to figure it out
and then uh yeah then it was fine uh
Python's pretty close to R so it's okay
not that's not the biggest deal in the
world um and then so I I interned uh
with John for for two summers uh it
great he was such a great like
internship intern manager and then um as
I was doing that I started um like
Contracting for this nonprofit that's
called called protect democracy um they
did a lot of very cool very important
work and it was very related to a bunch
of research that I had done um as an
undergrad and so I like my python was
good but it wasn't like it was like a
data scientist python it was like not
like a z Engineers python um so
everything was like everything I wrote
was messy I like barely wrote any
functions it was just in Jupiter it was
like okay like is not like you're not
Building Systems like you're just
building like kind of scripts in Jupiter
or something
so then um uh Sam uh this guy named Sam
Royston um he uh was the kind of like
head engineer on um on protect democracy
and uh he basically spent like we would
just pair program together and he knew
python unbelievably well um yeah and so
then like I would just we would just sit
there and then I would like watch him
code and I'd be like wow this is amazing
look at thisy
this guy this guy's really coding like
that's a real thing um and then we
switch and I would have to code and i'
be like I don't know how to do anything
and You' be like you should do it this
way oh no you should you should name
your variable differently oh actually
python can do it this way oh well this
is you know what a Dunder is and like
all the other you just like sat there
and taught me for like a summer um and
so between all of that then that's how I
learned python I don't know that's too
long but it was I remember freaking out
over the phone because I was like I want
to be a data science I gotta Learn
Python I only know R and you did it in
only three weeks right
that's after three weeks
but yeah know kind of how to do some
stuff yeah very inspirational yeah yeah
yeah scary I mean yeah and you know like
and then and then you over the years you
mentioned about like your experience
doing the pair programming and how you
enhance your programming and then now
that you're you're working full-time as
a data scientist at meta at streamlet at
Snowflake and you know more and more
experience under your reps reps is a
great one and the other thing is that
keeping your feedback loops really tight
so like uh we have a a a data scientist
engineer on our team his name is Zachary
um and he's just really he's like a
really good programmer and so I get to
hang out with him a lot and he sees a
lot of my code and then he'll say oh I
think maybe you should do this a little
bit differently or maybe you should do
this a little bit differently nicest guy
and so it's just like super helpful to
have him reviewing the stuff that I'm
working on and getting that good
feedback from people that know more
about python than you um and so reps
plus feedback equal you're G to get
better at it like you're just definitely
going to get better at I don't think
there's any way that you get worse at
something by doing it more having
feedback from people that are better
than you like I cannot imagine getting
worse at something having those two
things whether that's at work or whether
that's online like posting more stuff
online and getting people to look at it
and review it or like however that is um
especially if you're starting out with
python almost everyone that codes in
Python is better than you so it's not
that hard to find someone that that will
at least look at it and be like oh is
this good or is this bad you know should
I do it this way should I do it another
way so that's kind of the the gist of it
and I think that's also like the thing I
should have sought as when when I was
younger is you know I I I stumbled into
to places where I got that good feedback
but like seeking out quality mentorship
from not like oh yeah I meet with this
Mentor once a month and he's this super
senior person at this company and
they're a mentor like that mentorship is
fine and dandy but like the mentorship
of like I know this person they're
willing to sit down with me right next
to me and we can code on something like
that is that's a mentorship that is like
really would have been useful I'm
getting a lot more of that um you know
when I was a bit younger yeah yeah super
amazing would you have any advice for
the listeners like how could one find a
mentor oh uh well I guess if
you're I've never asked someone to
Mentor me and it's gone
well um so I
think I don't know if that's the
greatest way of doing it and I seen
people that try to do it that way and I
don't see it really succeeding that well
I think a lot of it
is it's got to be two-way right like
these people helped me in part because
they knew I could help them eventually
too like I showed that I was Earnest I
was intellectually curious and I wanted
to do well at an internship or help them
with this job or something and so you
you want to find
like you you want to find win-win
situations with other people that are
more experienced with you and so I think
the only advice that I have is it's it's
probably not going to go super well if
you just email random people who you've
never talked to and just say will you be
my mentor please like I I just my only
advice is uh that hasn't worked out for
me but what it has worked out for me is
hey these people are willing to invest
in someone in part because they want to
do something good in part because they
also want to get something out of it um
and finding more of those situations is
um is is key at least it's been key for
me but that said like who knows like I
I'm a I'm an N of one um I don't know if
my if my experience is actually the best
way of doing things I don't know if my
experience is you know reflective of
other people I can just tell you you
know my story and if that helps um then
great but if it doesn't then
sorry yeah super super helpful uh indeed
and I think it will be very valuable for
the listeners um yeah
so I think we're we're bracing through
our questions um and yeah recently You
released this amazing tra component
called STP wall um could you tell us a
little bit about it and like your your
thought process of how you actually you
know came up with the idea and how did
you actually create the component yeah
so one thing that's kind of been in the
back of my head
for I mean at least a year if not longer
is I don't understand why there aren't a
bunch of data scientists that are making
money online with data science projects
like I see software Engineers do it
right like every every other week on
Twitter I'm sitting there and there's
some software engineer out there that is
like I have made two ,000 this week with
my cool little app or something and
they're like I just hit an MR of like
$5,000 that's awesome that's so great
why do I never see a data scientist
doing that like and so there's a few
options in my mind either data science
is not that valuable right and
possibility two um data science is only
valuable inside companies that spit off
lots of data and so doing it outside of
companies is kind of impossible um or
three there's like a skill issue um or a
capacity issue where it's like oh well
data scientists could be able to do this
but they're missing something um and so
I thought I and I still think that
streamlet is a unique and excellent
Avenue for this type of work um for a
cool like you make some cool large
language model prompt but you don't want
to sell that prompt or you just want to
sell access to it you know so if you can
do that in streamlet when I pay when I
charge people like you know 10 bucks a
month or say hey give me 20 bucks and
I'll give you access to this app so I
tried to I tried to do that in streamlet
and I realized oh actually that's pretty
hard that that's that's actually really
hard to do because you have to do
authentication right you have to do all
the connections with the different
payment providers you have to like
depending on how much you're charging
and what you're doing you might have to
set up like like an actual business you
know like with stripe or something like
use like stripe Atlas or or whatever
else like these things are it's not
trivial it's definitely not trivial so I
created SC payall as a hey let's just
see if
people like let's see if I can break
down some of these hypotheses and let's
see if I can make it a little bit easier
for people to collect money online for
their work um my goal is that I want to
see lots of streamlet developers that
are making either a full-time living or
a parttime living with the cool work
that they're doing right like I want it
to be like the App Store where you see
all these Indie developers that create a
cool little app like you remember those
those those initial iPhone apps when
like I didn't even have an iPhone I had
an ey touch right and they had the
little you know the the dumb apps like
there was some there was one app that
went viral that was like oh just buy
this app to prove to other people that
you're rich and they just charged like
$10,000 for your app for for the app or
something crazy or like $1,000 for the
app and they made so much money or
another silly little app that was like
oh this app will help you like it just
like is a fake beer drink and it just
had a nice cool animation when right
like you remember those things or like
or like all these other random like
iPhone apps like I want that for data
science where people can make cool
little silly things and then make some
money off it and then eventually not
silly things like uber Uber is on the
App Store like it is the app right so
like you need that your Facebook or
Airbnb or like other things like these
are apps first airb probably less so but
Uber yes Facebook yes Facebook exists on
the App Store um so I I like want and
that's like a very high aspiration
that's pretty ambitious for for what
what streamlet is but in order to do
that we kind of like I wanted it to make
it easier for people to accept money
online that's like suggest it um and so
SC pay wall is a way where you can like
uh basically use this Library add a
bunch um at these you know a couple
little calls that will um do Google
authentic
um and then the second thing that it'll
do is um it'll uh it'll check if the
user that uh you know signed in with
their Google account if they're one of
your subscribers onrie if they paid for
um paid for this product already and if
they're not then it prompts them to um
pay for it and if they don't pay for it
then they can't see it so that's kind of
the uh the G fit it's not the most
complicated thing in the whole world um
but yeah I I I test it out um I think it
works pretty well and it's just like
something that has been like scratching
at the back of my brain for like years
now so I know I'll have more um products
and more thoughts on this in the future
but this is kind of a first sitation of
it yeah super super cool um and yeah
definitely um I was also wondering um
like a year or two ago and then I think
I asked over on Twitter like does anyone
know how we could connect our stret app
is there any component available and at
the time there were none um and yeah
here we are today and it's it's super
exciting to see that and how you could
help to unlock you know new avenues of
content creators triet app creators to
monetized and maybe make a living out of
it right hopefully yeah or just make $10
like just making $10 online is addictive
like as soon as you make $10 you're like
all right I'm gonna want to make a
hundred like right that point be really
cool right right right definitely and
yeah so we've reached our last question
which is like
what are some advice that you would like
to give to your younger
self
um I think the so this is probably
outside of streamlet but the one thing
that I value in people a lot are two
things are curiosity and kindness and I
would tell myself to be more Curious and
be more kind um those are the two
biggest things and so streamlet is a way
where I embrace my curiosity about the
world about the people around me um and
if I do that in a kind way then I think
uh I think I think that's the advice
that I would give my younger
self uh yeah so it's it's a big pleasure
to have you on the podcast and and this
is not the first time right it's the
second time and yeah I do appreciate
this and hopefully we could have you uh
for you know subsequent times third time
fourth time fifth time who knows well
thanks for having me on I really
appreciate it it's always so fun um to
talk with you you know now here in the
podcast at work whatever so I'm sure on
Twitter who knows right right right
right it's thank you so much yeah and
we're amazing to you know in the last
time last iteration we were um you know
internet friends and now we're we're
actually you know like working in the
same company yeah right so uh yeah who
knows what's gonna happen before uh
before the third time on this podcast
something crazy I assume yeah cool yeah
and so yeah uh would you like to um tell
the listeners how you they could reach
you uh yeah so I'm just Tyler J Richards
on Twitter uh that's probably the
easiest place to to reach me um we'll
put uh links to all the stuff that we
talked about uh in the description or in
the comments or somewhere uh I'm not a
YouTuber I don't know like well we'll
figure it out and then uh yeah the best
place to find me is on on Twitter you
can find my email there um my email is
just my username on Twitter gmail.com
that's a great place to um if you want
to chat about anything and then there
are other social platforms but I'm I'm
mostly a Twitter guy
so that's AIT uh we'll link the book um
I felt kind of weird about uh you know
accepting money uh while while being for
for a book while being also a this is
just a weird thing about me not anything
else but we just going to donate the the
proceeds of the book to ladies um so
that's the the gist of that um yeah and
that's uh that's me that's uh that's all
the stuff here yeah the super amazing
and yeah I I we will provide the links
to all of the social platforms that
Tyler is on and also the link to his
book yeah so do definitely check it out
perfect thank you so much yeah my
pleasure
Ver Más Videos Relacionados
Crie dashboards incríveis usando PYTHON, STREAMLIT e CHATGPT
How to Use Llama 3 with PandasAI and Ollama Locally
The New Youngest Self-Made Billionaire In The World Is A 25-Year-Old College Dropout | Forbes
I Tried Adding Google Auth To a Python Web App | ft. Streamlit
How I Became A Data Scientist (No CS Degree, No Bootcamp)
YouTube or TikTok Monetization: Which Platform Pays MORE? 💰
5.0 / 5 (0 votes)