Working at HuggingFace🤗 A chat with People Ops evangelist Emily Witko

Summary

In this insightful conversation, Emily Whitco, the People Operations lead at Hugging Face, shares her fascinating journey into the world of open-source AI. She delves into the importance of psychological safety within engineering teams, the challenges of knowledge sharing across technical and non-technical domains, and the unique decentralized culture at Hugging Face that fosters creativity and impact. Emily also discusses her personal projects aimed at promoting inclusivity and exploring pay equity using Hugging Face's cutting-edge tools. Throughout the interview, her passion for empowering people and fostering a supportive environment shines, offering valuable insights into building thriving teams in the rapidly evolving AI landscape.

Takeaways

• 😀 Emily Whitco works in a generalist people role at Hugging Face, focusing on employee engagement, recruiting, and people operations.
• 🤖 Despite not having a technical background initially, Emily learned about Hugging Face's work and felt drawn to their mission of responsible AI development.
• 💻 Hugging Face follows a decentralized approach, where employees set their own goals and projects based on their interests and skills.
• 🌱 Emily helped create structured impact plans for teams to provide guidance and clarity for new hires on their expected contributions.
• 🤝 Psychological safety and inclusivity are highly valued at Hugging Face, with efforts to promote a supportive and empathetic work culture.
• 🌐 Knowledge sharing at Hugging Face happens through asynchronous communication, ensuring conversations are documented for easy reference.
• 🌈 Emily has been experimenting with projects related to pay equity analysis and inclusive language in job descriptions using Hugging Face's tools.
• 📚 Keeping up with the rapidly evolving AI landscape can be overwhelming, so Emily recommends selectively following trusted sources and contributors.
• 🌇 Emily values the openness and focus on potential at Hugging Face, where backgrounds are diverse, and growth is encouraged.
• 🚀 Hugging Face aims to be the platform for AI builders, with a kind ambition to achieve success while prioritizing its people's well-being.

Q & A

What is Emily's role at Hugging Face?

Emily's role at Hugging Face is described as a very generalist people role, focused on tasks related to people operations, such as employee engagement, recruiting, and hiring.

How did Emily end up at Hugging Face?

Emily initially discovered Hugging Face while researching chatbots for a banking startup she was working at. She was impressed by Hugging Face's open-source leanings and responsible approach to AI, and eventually applied for a role there after discussing her interests with the CEO.

What is the culture like at Hugging Face regarding decision-making and goal-setting?

Hugging Face has a decentralized approach where teams and individuals set their own goals and projects, rather than a top-down cascading structure. This promotes autonomy and a lack of rigid hierarchies.

How does Hugging Face foster psychological safety among its employees?

Hugging Face has a team dedicated to maintaining community standards and monitoring the platform. Internally, they have small, project-based teams with team leads responsible for ensuring team members feel validated and supported. The company also values kindness and ambition, promoting a supportive environment.

What project is Emily most proud of at Hugging Face?

Emily is proud of creating structured impact plans for each team, which provide new hires with clear goals and projects to work on, reducing anxiety and increasing confidence.

How does Hugging Face promote knowledge sharing among its employees?

Hugging Face encourages asynchronous communication over meetings, which means all conversations are written down and can be easily referred back to. Employees are also encouraged to share their work on social media and follow their colleagues' updates.

How has Emily been experimenting with Hugging Face's tools?

Emily has been working on two projects: one related to pay equity, using Hugging Face's data and models to ensure fair compensation across different countries; and another tool using Hugging Chat to analyze job descriptions and social media posts for inclusivity.

Who would Emily be interested in interviewing in the AI space?

Emily expressed interest in interviewing Timnit Gebru, who previously worked at Google and is now working on ethical AI. Emily admires Gebru's confidence in standing up to major players in the industry and speaking up for what's important.

How does Emily manage staying up-to-date with AI news and developments?

Emily recommends following a limited number of trusted open-source contributors and projects on platforms like Twitter, GitHub, and the Hugging Face Hub and Discord. She also learns a lot from her colleagues at Hugging Face, who share their work regularly.

What advice would Emily give to someone feeling overwhelmed by the constant flow of AI news and developments?

Emily suggests being selective about the pieces of AI and open-source corners one engages with, as there is so much opportunity that it can be exhausting to try to keep up with everything. Picking favorite contributors and limiting the number of sources is key.

Outlines

00:00

Renee's Introduction and Emily's Background

Renee introduces the podcast and Emily Whitcomb, who works in people operations at Hugging Face. Emily shares her background, starting from customer-facing roles and transitioning to a more technical role at a banking startup before joining Hugging Face. She explains how she discovered Hugging Face while researching chatbots and was drawn to their open-source approach and ethics team, leading her to apply for and eventually accept a role in people operations.

05:02

Emily's Journey to Hugging Face

Emily recounts her conversations with the CEO of Hugging Face, Clément Delangue, during the interview process. She was initially considered for a customer success role but was offered a position in people operations instead. Clément emphasized Hugging Face's culture of valuing potential over past accomplishments, which resonated with Emily. She also discusses the ambiguity of her role and the company's decentralized approach, where employees have the freedom to choose what they work on.

10:03

Open-Source and Decentralized Culture

Emily explains how working in an open-source community and at Hugging Face requires flexibility, confidence, and an experimental mindset. She highlights the decentralized nature of Hugging Face, where small teams set their own goals and projects, fostering a sense of ownership and impact. Emily also touches on the company's value of "kind ambition," balancing ambitious goals with empathy and thoughtfulness.

15:03

Psychological Safety and Inclusivity

The conversation shifts to the importance of psychological safety, particularly in the open-source world, which can sometimes lack inclusiveness. Emily discusses Hugging Face's efforts to create a safe environment internally, including having community standards, hosting events for underrepresented groups, and fostering a flat team structure with team leads focused on ensuring team members feel validated and supported.

20:04

Creating Impact Plans and Onboarding

Emily discusses one of her proudest achievements at Hugging Face - creating structured impact plans for each team to help new hires understand their goals and expected areas of impact. This initiative aimed to address anxiety and uncertainty that new employees sometimes experienced in the company's decentralized environment. Emily emphasizes the importance of setting expectations and providing guidance while maintaining the company's agile and open culture.

25:06

Knowledge Sharing and Asynchronous Communication

Emily shares insights on how knowledge sharing and technical explanations are valued at Hugging Face. Engineers are expected to communicate complex concepts clearly and without jargon to non-technical team members. The company's policy of prioritizing asynchronous communication over meetings ensures that all conversations are documented, creating a rich knowledge base that employees can access at any time. Emily also highlights the supportive culture where asking questions is encouraged.

30:07

Experimenting with AI Tools and Models

Emily discusses her recent projects involving experimenting with Hugging Face's AI tools and models. One project focuses on pay equity, exploring how to ensure fair compensation across the company's global workforce using data sets and models. The other project involves creating an inclusivity tool that analyzes job descriptions and social media posts for potential biases and suggests more inclusive language.

35:08

Keeping Up with AI News and Developments

Emily acknowledges the challenge of keeping up with the constant stream of AI news and developments, both within Hugging Face and in the broader AI community. She advises being selective about the sources and channels one follows, suggesting focusing on favorite open-source contributors and platforms like Twitter, GitHub, the Hugging Face Hub, and Discord. Emily also highlights the privilege of learning from her colleagues, who share their work and insights across various social media platforms.

40:09

Closing Thoughts and Future Events

The conversation concludes with Emily sharing her availability on LinkedIn for those interested in keeping up with her work. She mentions upcoming events in Sacramento, California, and Europe in July, although no specific details are provided. Renee thanks Emily for the insightful discussion and encourages listeners to engage with the podcast and guests.

Keywords

💡Psychological Safety

Psychological safety refers to a work environment where individuals feel safe to take risks, voice their opinions, and express their thoughts without fear of punishment or ridicule. This concept is crucial for fostering innovation and collaboration within teams, especially in technical fields like engineering. In the script, psychological safety is discussed in the context of engineering teams and remote work, highlighting its importance for employee engagement, team dynamics, and overall productivity. Examples include creating a culture where employees are encouraged to share ideas and feedback openly, contributing to a more inclusive and dynamic workplace.

💡People Operations

People Operations, often referred to as 'People Ops,' is a modern take on human resources that focuses on optimizing employee performance and enhancing their work experience. It encompasses traditional HR functions like hiring and payroll but with a stronger emphasis on culture, engagement, and development. In the script, Emily Whitco describes her role in People Operations at Hugging Face as a generalist role focused on various aspects of employee well-being and organizational development. Her approach includes focusing on her favorite aspects of the role and whatever needs to get done, illustrating the flexible and dynamic nature of People Operations in startups.

💡Remote Work

Remote work refers to the practice of employees working from a location outside the traditional office environment, utilizing digital communication tools to perform their duties. The script discusses remote work in the context of its impact on psychological safety and team dynamics. It highlights the challenges and opportunities remote work presents for maintaining team cohesion and ensuring that employees feel supported and engaged, regardless of their physical location. Examples from the script include discussions on how remote work has influenced the operations and culture at Hugging Face.

💡Open Source

Open source refers to software for which the original source code is made freely available and may be redistributed and modified. It represents a collaborative approach to software development that encourages participation and contribution from a diverse community of developers. In the script, the conversation touches on the role of open source in Hugging Face's operations and the importance of contributing to and participating in open source projects. This context highlights the value of open source in promoting innovation, collaboration, and transparency in the tech industry.

💡Decentralization

Decentralization in an organizational context refers to the distribution of decision-making powers and operations away from a central authoritative location or group. In the script, decentralization is discussed as a core principle at Hugging Face, influencing its culture and operations. The company operates with a flat structure where teams and individuals are empowered to set their own goals and undertake projects independently. This approach fosters autonomy, innovation, and a sense of ownership among employees, as exemplified by the practice of not having cascading goal-setting processes.

💡Technical Background

A technical background refers to having education, training, or experience in technical fields, such as engineering, computer science, or information technology. The script explores the transition of Emily Whitco from a non-technical background to working in a technical role at Hugging Face. This transition underscores the inclusivity and diverse skill sets valued in the tech industry, especially in companies like Hugging Face that prioritize potential and learning ability over specific technical expertise.

💡Startup Culture

Startup culture refers to the norms, values, and practices that characterize the working environment of startup companies. It often emphasizes agility, innovation, collaboration, and a flat hierarchy. The script delves into aspects of startup culture as experienced by Emily Whitco at Hugging Face, highlighting the adaptability required, the generalist roles often embraced, and the dynamic nature of working in a startup. This culture is contrasted with more traditional corporate environments, showing how startup culture fosters a more flexible and innovative approach to work.

💡Inclusivity

Inclusivity in the workplace refers to practices and policies designed to create an environment where all individuals feel valued, respected, and supported, regardless of their background or identity. The script mentions efforts at Hugging Face to ensure inclusivity, such as hosting events for women in machine learning and developing tools to assess the inclusiveness of job descriptions. These examples illustrate the company's commitment to diversity and creating a welcoming and supportive space for all employees and community members.

💡Career Transition

Career transition refers to the process of moving from one job or career path to another. In the script, Emily Whitco discusses her transition from customer-facing roles to a more focused role in People Operations at Hugging Face. This transition highlights the fluidity of career paths in the modern workplace, especially in the tech industry, where companies value diverse experiences and skills. It also emphasizes the importance of openness to new experiences and the ability to adapt to different roles and responsibilities.

💡Impact

Impact in a professional context refers to the tangible effects or contributions an individual or team makes toward achieving the organization's goals. The script explores the concept of impact at Hugging Face, particularly how it guides performance reviews and project selection. By encouraging employees to focus on impactful work, the company promotes a culture of purpose-driven effort and meaningful contributions. This approach empowers employees to prioritize projects and tasks that have a significant and positive effect on the company and its mission.

Highlights

Emily Whitco discusses transforming people operations at Hugging Face and the importance of psychological safety in engineering teams.

Whitco shares her journey from customer-facing roles to a people operations generalist in tech, highlighting her transition to Hugging Face.

The significance of open-source platforms and the ethical considerations in AI are explored through Whitco's experiences.

Whitco emphasizes the value of a startup culture that prioritizes potential over past achievements.

The conversation delves into the concept of decentralization at Hugging Face and its impact on work culture and productivity.

Discussion on the role of team leads in fostering a psychologically safe and inclusive work environment.

Whitco highlights Hugging Face's commitment to ethical AI and inclusivity in the tech community.

Exploring the challenges and learning curves of working in an open-source environment and the importance of flexibility.

The impact of structured impact plans on employee engagement and clarity in roles at Hugging Face.

Whitco shares insights on effective knowledge sharing and the importance of clear, jargon-free communication in technical teams.

The role of asynchronous communication in enhancing productivity and knowledge accessibility at Hugging Face.

A discussion on the challenges of keeping up with AI advancements and the importance of selective learning.

Whitco expresses admiration for Timnit Gebru and her work in ethical AI.

Exploration of pay equity and inclusivity tools being developed by Whitco using Hugging Face's technologies.

The personal and professional growth opportunities provided by Hugging Face's unique work culture.

Transcripts

Renee here at unsupervised learning your

easy listening podcast for bleeding edge

open source Tech this episode sees me

speaking to Emily whitco someone who is

transforming the people operations at

hugging face has experienced a general

assembly and also alternate Futures and

sits on their board of directors it was

an excellent conversation on

psychological safety in engineering

teams and remote work hope you enjoy

hello I do the introductions afterwards

so that you don't to sit here and um

hear how wonderful you are

but you are the what what do you have

like a like a title like it's like

people operations or it's a great

question the the short answer is not

really um it's a very I I I describe my

role as being just a very start upy sort

of generalist uh people role where uh

you know I focus on

uh a couple of my favorite things and

then also whatever needs to get done and

so I have a very very people generalist

role yeah and has it changed a lot like

I imagine it has but I wanted to talk

about like your because previously were

you at very Tech like do you have a very

technical background or any exposure to

much software or like yeah it's not

really a question it's more of a run on

sence

but no I I really do not actually um so

uh I'm trying to think of how to tell

this story in a shortish way but most of

my career has actually been spent in

like customer facing roles so account

management um like Enterprise account

work um and has not fully been focused

on software so so a chunk of my career

was at at uh General Assembly which is

sort of a a tech education startup that

I feel most people have heard of um and

then I'd say uh my most technical job

previous to hugging face was at a little

teeny tiny six-person

banking startup that I was working at

for a couple of years um where we

actually were building software and

because it was such a tiny team I was

not writing the code but I was involved

in that software in essentially every

other way that you could possibly think

of um in addition to working with our

client or working with our customers so

that was sort of my first really foray

into understanding the complexities of

software um but at the end of the day we

were still building essentially just an

app right we're all fairly familiar with

apps at this point and and particularly

banking apps they're not the world's

most Cutting Edge piece of technology

all the time and so when I moved to

hugging face it was

definitely a period of transition that

that sometimes I still feel like I'm in

where all of a sudden these technical

conversations were very different than

the technical conversations I've had in

the past um and much more much more

complicated I want to get into like how

did how did that move happen like how

did you end up at hugging face yes so

when I was working at this banking

startup we actually were starting to

think about what an automated chatbot

could look like and so I had done some

searching around about some of the

easiest ways for us to build our own or

or possibly even some of the ways that I

could build a chatbot without using some

of our very limited engineering

resources at that uh startup and that's

actually how I found hugging face um and

and I started digging through the

website a little bit um and pretty early

on in trying to explore hugging face

realized that it's a very technical

place right even for folks that end up

at the hugging face website it there's

it may even be difficult to know what

hugging face is once you've landed on

our homepage if you're not an engineer

um and so I just was doing a bunch of

research I think digging around on like

edit or GitHub or looking through the

documentation just to try to learn a

little bit more about what hugging face

was and as I learned a little bit more

about like our the open- source leanings

and the fact that there is an Ethics

team here and the fact that they're

thinking about AI in a really

responsible way I started to get more

and more interested in the

organization so I applied to an open

head of customer success role that was

at uh on the job site at the time and

very quickly heard from the CEO clal who

asked if we could chat and so we had a

couple of conversations that were really

quite informative about the culture at

hugging Bas and we talked about my

experience and then after our second is

conversation he said I think we really

need someone with technical skills in

this customer success role will probably

be looking to hire an engineer um but

what I here is that you really love to

build teams you really like to think

about Employee Engagement you like to

recruit and hire and source and all of

those things would you be interested in

joining the team and doing those things

and my reaction was I was super excited

um it had been a switch that I had been

considering making for a few years um

but wasn't really sure how to shift from

the customer service space into the

people space um and more or less on the

call

said I'm I'm excited by this Prospect

but I just need you to know that this is

not work that I've ever done before in a

professional capacity um and his

reaction was was what sealed the deal

for me and and why I really wanted to

work at hugging face was and it was

essentially like I don't I don't care

we're not particularly interested in

what people have done before we're we

much more interested in what people can

accomplish once they joined us um and it

was really just

that I don't know that that that ethos

that really got me excited to work for

hugging face and and he did further

clarify his I don't care

to be a little bit less harsh very

French um and and described that you

know we have physicists that are now

working as machine learning engineers

and we have linguists and philosophers

and just the the background of the folks

at hugging face is um or backgrounds I

should say are completely varied um

which a makes it a really interesting

place to work but

B is a is a culture that I always wanted

to be a part of right I I was excited by

the idea that um we're more excited

about people's potential than we are

essentially about what they've

accomplished in the past or what school

they went to or exactly what their

resume says so uh yeah that's a

long-winded way of saying that's how I

ended up

here that's a great story yeah I I was

really I was really impressed with the

leadership

um and the last part of it I'll say

which sort of relates to the fact that I

don't really have a a specific job title

is my partner is a lawyer here in the

states and I remember he was really

nervous when I signed my paperwork

because I didn't have a job title and I

didn't really even have a job

description and I can remember a

conversation we had where he was like

how do you know what you'll be doing and

my reaction was just sort of like I

don't but we'll figure it out along the

way and so um it's been just over two

years of my time at hugging face and you

know my role I think has has shifted

with sort of company needs but um it's

been it's been a really wonderful

learning experience that leaves me in

like a couple paths to go down but when

you mentioned about it's it's not uh

it's not like a Laz Fair attitude but

it's more like um kind of an openness

like you said like being like oh are you

generally like that would you say in

terms of like career-wise or are you um

have you found that that's been

something that people need to to kind of

adopt to to work in open source is

having

a uh a more experimental mindset I

guess that's super interesting um I know

for me personally I get excited by new

things and so anytime I'm given the

opportunity to like try out a new role

or or think about even just like a new

project idea um um I'm definitely the

type of person to say yes and and Dive

Right In um and so I think

uh that's part of why sort of the the

ambiguity of the space doesn't scare me

is you know I'll I'm excited by it in

some ways

um I guess I feel like there

are possibly some some similarities

between working in a startup environment

and engaging in open- Source Community

spaces prob mostly in just the sense

that you never know what's going to

happen next or you never know what sort

of uh comment somebody will make or

you'll never know what sort of

improvement you'll see in the project

that you're working on um and so it does

require I would say a significant level

of of

flexibility um and probably along with

that um a level of confidence in your

own

skills and and being able to sort of

step into a different scenario and and

and feel confident enough to stay

involved um and so I I guess I mean I

never really thought about it before how

you might be able to extrapolate it to

being involved in open source

communities but um it is an interesting

it is an interesting thought I the the

kind of context for that is I found it

very despite only having worked in

startups I found it very different uh I

think I think it was a almost an ego

activity stepping into open source being

like okay it's constant feedback right

from the community and it's like the

awareness that um you don't necessarily

even if you own a function it's like

it's not yours you're doing it for the

betterment of whatever so you need to be

you can't be I don't want to say

pigheaded but I'm going to say it you

can't be pigheaded about decision making

and it's so you posted on LinkedIn about

decision- making at hugging face and how

that like I guess I'm asking like was it

just an intuitive like a duct to water

type thing with you where you're like

yeah this is cool this is fine because

that's that's the vibe I'm picking up

where you're like you're just the kind

of person that just goes with it or was

it uh was it a little bit of a learning

curve and if it was fine have you seen

people that do go through that journey

and what does that look like and what do

they need to kind of learn to

overcome their own

issues yeah no I mean even for me uh it

definitely was a learning curve I think

uh both both being part of an open

source community and and experiencing

the things that you described and also

working for a place like hugging face

that operates internally very similar to

I would say the way that open source

communities work by Design right so the

word that we use at hugging face over

and over is

decentralization I think your experience

with open source is a good example of

decentralization the example that I use

for folks um that I feel like especially

those that are coming from more

traditional startup environments are is

that um we don't have like a

cascading goal setting process each year

right like RC CEO is not setting goals

or okrs for the whole company and then

each team is sort of building off of

that it's completely flipped right so by

Design the folks that are doing the work

the folks that are involved in our

communities um are setting their own

goals and not only that they're coming

up with their own projects a lot of the

time and

so it's it's it's interesting

because you can imagine interviewing an

engineer a candidate who wants to work

at hugging face and saying to

them no one's going to tell you what to

do and you can choose what you want to

work on and you can imagine that some

folks that's a dream right some

Engineers are like sweet like I don't

have to talk to anyone and I can just

work on what excites me um we love when

that happens for other people including

myself it's a little bit of a learning

curve because all of a

sudden it's it's it's not only that

you're sort of left to your own devices

it's just that all of a sudden you have

a level of responsibility that you may

not have had at your other organizations

right there's sort of a lack of of top

down direction that can sometimes leave

you feeling a little bit like you're

floating in the middle of the sea and I

feel like I've heard similar experiences

especially from folks who are trying to

get involved in open Source projects is

almost that um What's the phrase I'm

trying to think of where it's just like

that Paradox of choice where you have so

many different projects that you can get

involved in yeah that you never start

right or you never pick one like

analysis paralysis kind of yes yeah

exactly um and so sometimes working at

hugging face can feel the same way yeah

because there's a thousand projects that

you can be involved in both internally

and with our communities um um and so

that's definitely part of like a growing

pain here at hugging face I would say

and something that um I had to learn was

to to limit all those projects that I

get really excited about and really come

up with a focus for myself um so that

the work that I was doing the work that

I am doing is highly impactful

um and and thoughtfully done yeah nice I

wanted to speak a little bit more about

like the work that you're doing but it

led me to think abouty like

psychological safety because I think

that that's maybe one of the foundations

of impactful teams I say this not having

been in people Ops very much at all but

you know I'm I'm like one of those

LinkedIn people that just say all these

things and it's like what qualifications

do you have and it's like well I've got

an opinion so but I I do feel that

wherever you work psychological safety

is so important but even the idea of

saying that it's important it puts you a

little bit uh in the firing line if

you're working somewhere that isn't

psychologically safe right because then

you're looking like a snowflake so it's

kind of something that I wanted to to

touch on like because in my experience

I'm I'm stereotyping here but I'm of the

opinion that software engineers and

Engineers ml Engineers would be like I

just want to do the work I don't need to

worry but it's like you do have the type

of people that do need that constant

reassurance and checking kind of thing

so like what does that look like at

hugging face if there's no top down

culture like what kind of things do you

do in team building that kind of thing I

understand my questions are always four

questions in

one we're getting better at

that it's a very thoughtful question um

but but in a similar vein I feel like I

have four answers so so to start what I

will say is I I don't need to tell you

this um

but the like quote unquote world of Open

Source is not always the most

psychologically safe place right

um it's the

internet we all know what happens

sometimes on the internet and especially

from a place of inclus

iess um we we don't have exact figures

right and I I feel like the numbers that

I use when I talk about this are from an

old um maybe three or four year old

GitHub study but open source

contributors that are women we're

looking at essentially fewer than

10% um and there are I'm sure you know

this these are complicated conversations

and I'm sure there are a variety of

reasons for that but it's not always a

psychologically safe space and folks

that face systemic biases or or have

barriers in their life are less likely

to engage in places that are less

psychologically safe and so I think that

that that complicates communities that

exist on the internet where people feel

like they can behave in some less than

appropriate ways um internally at

hugging face we spend a lot of time

um making sure that our employees do

feel safe um we do it I I would say in a

few ways one

um I was gonna say is sort of your a a a

more standard or or traditional approach

um but I feel like using those words

sort of undercut the amount of work that

we've done to do it correctly but we we

have a team of folks that um is

incredibly thoughtful and works very

hard on our community standards um and

those are those are public and those are

things that I can share with you and and

they are you know our Hub is monitored

by both our team and the community um

and we you know don't tolerate certain

behaviors in our community which um is

something I'm trying to share a little

bit more widely with folks who who may

not come currently exist in these

communities um for example just maybe

two weeks ago while I was in Paris we

hosted a women in machine learning and

data science event I was gonna get to

that um that was one of the that was one

of the things that I was trying to kind

of head in the direction of that yeah um

go ahead sorry no and all I was gonna

say is is like talking to groups you

know of like those and encouraging in

open- Source contributions if folks you

know if folks in the women in data

science group are not already

contributing to open source projects on

hugging face or wherever to try to

encourage folks to to do that and

explain that we generally you know we

try very hard to have a psychologically

safe

platform um internally at hugging face

uh we our team is is very very flat

right and that's part of the

decentralization and that's part of why

we don't have this sort of pyramid

shaped

organization and what that means is that

we tend to have fairly small teams that

are very Project based and each of these

teams consists of maybe between four to

eight people probably not much larger

than that um and there tends to be a

team lead for each project um and I say

this because our team leads serve an

important function a they're they're

probably the point person for other

people at the company to ask a bunch of

questions of but B they work as that

like quote unquote leadership to make

sure that the rest of the team is

feeling validated is feeling safe and

also um is part of like uh feedback

cycles and making sure that folks are

getting uh yeah support and information

and setting goals together when they

need to um we have a one of our values

at hugging face is um to have a kind

ambition and what that means is that

we're like shooting for the stars in

terms of hugging face success you know

we want 10,000 employees and we want to

be the platform for AI Builders um but

at the same time we don't want to take

it like take work too seriously we never

want to put work ahead of essentially

our people our our humanness um and I

think that most of our team uh takes

that truly to heart and is a very

supportive and and psychologically safe

group that

uh I don't know I this I don't mean for

this to sound like chastising or

anything but like that I'm so proud of

every day like I am I am very proud of

this team um for both being incredibly

smart people working on cutting edge

technology and doing it with with

empathy and

thoughtfulness um which I think is just

a really special

combination you've seen the team scale

from like how many people were when you

started I think it was about 40 when I

started and now we're we're close to 200

yeah wow that's a lot what's the like I

don't want to put you on the spot but

like what's do you have a an example of

like something that you've done at

hugging face that has

been very enjoyable for you like it

doesn't have to be like this was the

biggest thing it's just like what what

kind of did you enjoy doing the

most um good question we talk a lot we

talk a lot about impact at hug face um

we don't do standardized like 360 yearly

reviews performance reviews at hugging

face we we do them much more on like

project Cycles we sort of leave it up to

our teams to have discussions when

appropriate regarding performance um but

the word that is that you'll hear over

and over like cascading throughout the

company is the word

impact um

and when I started at hugging face and

was early on in the process there wasn't

a lot of definition behind the word

impact right so we would have new folks

join the team uh as a random example

someone would join uh our like front-end

web development team and our uh

onboarding materials got them set up

technically and they met one or two

people on the team and then they would

we would say okay go have an impact and

then just sort of send them off on their

on their way um and what what I found

was that there was a lot of um anxiety

is probably the best word for for new

Folks at hugging face because for the

first three months the first six months

even sometimes of not knowing exactly

what you're supposed to be doing and not

knowing where the goalposts are were

were did creates an anxiety and so I

feel like one of the one of the projects

that I'm most proud of and and really

did enjoy doing at hugging face was uh

creating like structure for impact plans

for each of our teams um now we are a

startup and we're super agile so our

goals look very different than sort of

traditional kpis right we they're not

hugely specific but each team now when

you hire somebody new you actually have

a document or a couple of paragraphs

that you can share and say here are our

goals here is some projects where you

can fit in um and these are spaces in

which we expect you to have an impact

and just setting those expectations

right away uh went a long way toward

making our new folks feel super

comfortable and excited to work at

hugging face as opposed to feeling like

uh they weren't aren't really sure where

they should be spending their time um

and I think it's actually been a very

powerful change in our culture is to

have some of this stuff documented um

and give people some confidence when

they start because you can have two

different kind of I see two different

types of people where it's like some

people without the kind of guard rails

it's like oh this is amazing this is so

freeing and other people it's like that

guard rail is a gift and for people like

me that guard rail is a gift because

otherwise I'm like I'm running in this

direction is it even the right direction

that's right

yeah my my kind of last question before

I do my my typical weirdo questions all

the other questions were normal um is

like do you have

any specific pointers on grabbing

those I imagine you've spoken to so many

technical people and being able to pull

out insights from them like you said

that there was you had to go digging in

documentation like how have you

found grabbing Knowledge from technical

teams or knowledge sharing I should say

building a culture of knowledge sharing

I know that documentation is important

for some folks but sure yeah um I just

had a conversation with one of my

colleagues while I was in the Paris

office because we were speaking with an

engineer who um was a recent graduate

and so someone who didn't have a lot of

work experience

and that person was asking me some

really technical questions about the

hugging face platform and I more or less

did the thing of being like I can't help

you but my colleague can and I pointed

this person I sort of pushed this person

in the direction of my colleague um but

then found out a little while later that

my colleague also couldn't figure out

what this person was asking um and so

that story is just leading me down the

path of saying

that at hugging face we think the best

marker of a good engineer particularly

somebody who works on really complicated

projects is someone who can very clearly

explain the work that they're doing to

non technical folks right you to be a

great engineer you should be able to

speak without jargon right you should be

able to speak in plain langu language um

and and have your work be accessible to

all different kinds of people right

that's not always the case um but I do

think it's something that our engineers

at hugging face are exceptional at um we

in a I feel like one of the ways that we

knowledge share really well at hugging

face and it it's kind of um part of our

culture that I think sometimes surprises

people is that we don't have any

meetings um it's

it's like one of our very very few

policies at hugging face is that we

don't we do really encourage

asynchronous communication over meetings

every time for a couple of reasons one

is that as you know with being in

Australia and me being in California the

number of times that we're available to

meet is fairly limited and we don't want

people waking up in the middle of the

night to take a meeting um but almost

more importantly is that when we don't

have meetings it means that all

conversations are written

down and so what that means is that I

can go back in our slack in Google Drive

at any given time and say you know what

I don't understand this feature on the

Hub for example I don't know what

hugging face assistants do what is this

and I can literally just in slack type

the word hugging chat assistant and go

find conversations that have happened in

the past few weeks months years whatever

um that would likely answer any question

that I have about it um and so the

asynchronous communication also means

that we essentially have like quote

unquote documentation in all different

kinds of forms um that

allow people at any given time to get

information that they may not have had

otherwise um yeah and the last thing

I'll say on that point is that it sort

of goes back

to the kind ambition piece is that I've

never had a situation where I've asked

for technical help from someone on the

team and felt

like they thought I was stupid yeah

there have definitely been situations

where my teammates have said listen I

don't have time right now you know you

can figure it out or we can do this next

week um but everyone has been so

generous with their

knowledge

um that I think I think that's something

that I tell new people at hugging face

also or even new contributors to our

community is it can sometimes be scary

to ask the question um but it's really

important to ask the question um and no

one's going to think you're stupid for

having asked

it but you sort of have to do that one

one two or three times in order to get

comfortable with it you just reminded me

about uh what was your

hugging hugging agent called

duck oh uh that's a good question what

did I call it you called it the rubber

duck buddy yeah the rubber duck buddy so

on because generally my my questions are

like if you could have invented anything

what would you invent but I'm

sandwiching the two together uh so I I

I'll sandwich that together with the the

the rubber duck have you

been experimenting a lot with creating

your own things using because I assume

that you get

into different functionalities like

within hugging face and on that post

that you did you were like hugging face

is different things to different people

say like what is it to

you yeah

um that's also a really thoughtful

question the answer the short answer is

yes um I've been trying to for for some

time now working on two separate things

that are sort of uh related um but one

is thinking

about uh pay equity and so thinking

about our our team here at hugging face

um and also you know tools that people

can

use in other organizations um

particularly where there are complicated

pay structures right and so we are so

lucky in that we have the ability to

hire globally and so we are currently

200 almost 200 people and we cover 30

countries and so it can be really

difficult to know whether or not you're

paying equitably when there are so many

variables including countries where

people are and so um

um I'm essentially have access to really

wonderful data data sets and different

models at hugging face that I've been

playing with for you know the past I

want to say six or eight months um as

someone who is not an engineer it takes

a while I think for someone who is an

engineer it takes a while but for

someone who is not it takes T times as

long um but it there really are an

amazing Treasure Trove of tools and

people who are willing to support you um

in the hugging face Community um

sorry and then the uh the last piece or

the another assistant that I made

recently a different assistant that I

made recently that I think has a more

immediate and probably wide ranging

applicability is uh just a really quick

tool using our hugging chat uh where it

thinks about inclusivity and so if

you're writing a social media post or if

you're going to post a job description

you can essentially paste it into this

model um

and in theory if it works well it will

tell you whether or not your your post

is inclusive and if not give you sort of

suggestions for how you can improve it a

little bit and and and do some of that

editing for you um that way we're not

having job descriptions that are fully

mailed coded or only thinking about

folks who have phds and those sorts of

different things so it is it's been a

lot of fun seeing some Cutting Edge

tools and thinking about ways that I can

play with them

uh in topics that are important to me

which is pretty cool that's that's the

whole aim of the game I think for

scaling adoption for everyone not just

the non technical people because I see

so much it's like this is amazing it's

like yes but how can I use

it what can it

do uh I generally ask like who would you

interview in the open source space or

the AI space it can be anyone they can

be dead

um they who would I

interview um I'm trying to think about

people I don't work with because that's

cheating I'm just so inspired by some of

the people that I work with which is

great um but I think if I could

interview someone in the AI space I

would be interested in interviewing Tim

nit GBU um who is uh the the part that

is less interesting is that she used to

work at at Google and and lost her job

several years ago and in sort of a

whistleblowing um moment that you may or

may not know about um but is working

very hard now in terms of ethical Ai and

making it a safe space and um I think

has a really direct approach to some

really hard questions and I would I

would love to ask I think probably more

people focused questions meaning meaning

sort of circling back to how we started

this is not how are you you so confident

in your platform but how did you grow

your confidence how do you feel able to

stand up against some of the biggest

players in the space and speak for what

is actually important um and not back

down which I think are are are really

important attributes and I I really look

up to up to her and for the work that

she's done I think for a lot of those

reasons

thank you uh my other thing is and I

feel like it's I'm going to relate it in

such like a Leap Frog way I'm gonna be

like yes totally totally related but

where do you do you experience I feel

like you would the kind of the feeling

of bomo like you have to keep up

constantly with all of the AI news and

everything like that and it's like well

where are you getting your

information like Emily wakes up in the

morning what are you looking at like

where do you go we talk about that a lot

at hugging face and partially because

even

without the whole world of AI and

without AI being so hot right now and

there being conversations

everywhere as I sort of mentioned

earlier we have a hundred projects

happening at hugging face at any given

time and so it's even I was going to say

difficult but I would say impossible to

keep track of everything that's

happening even within the company and

so we have an incredibly active instance

of slack with I don't even know for sure

probably 2,000 channels or something

that are all active and um we we tell

folks to really limit what channels they

join and uh how many convers ations to

be a part of because it can be really

exhausting um and so uh it it's sort of

a double-edged sword right you want to

know everything that's going on and yet

you could spend your whole day just

reading about what's happening in the

world of AI and not actually be able to

do anything else because there's so much

new stuff happening all the time and so

I think that folks have to be really

selective about the pieces of AI that

they're learning about or the you know

open- Source corners of the world that

they're engaging with because there's

just so much opportunity um that being

said I do have the privilege of learning

a lot from my colleagues um the way that

hugging face is set up is that we

actually also don't have a marketing

team we uh essentially ask that all of

our employees be their own marketers and

share on Twitter or on the Hub or on

LinkedIn what they're working on what's

been happening in the world um and that

means that just by following all of my

colleagues I get a really good sort of

General sense of what's happening in the

world of AI um and so I guess if folks

are asking sort of the similar question

of how do you keep up it say pick your

favorite open source contributors and

maybe limit it to 10

20 and and follow on Twitter um or

follow on GitHub or follow on the

hugging face Hub um we actually have

social posts now so that's probably a

good place for folks to start

also the Discord um yep yeah it's very

is also Super Active which is awesome I

think we're probably 60,000 people in

that in that Discord now um so there's

there's essentially pick your favorite

social platform and you you can learn

about what's happening in the world of

AI just don't pick too

many yeah I think that does come back to

um what you said about having uh

conviction and it's almost like having

that strong conviction in your own

beliefs and the idea that like no I'm

doing this like I'm not worried about

what other people think if they think

I'm behind like oh you're falling behind

bro it's like no I'm not bro I don't

care so it's like something about

confidence because I see a lot of I'm

saying younger people like I'm in my 30s

but like I see a lot of younger people

that feel like they have to be on Reddit

constantly and it's like I get those

feelings as well but um you're back in

California now do you travel a lot where

do people keep up with you LinkedIn

Twitter uh yeah LinkedIn is probably the

best um I do try to travel uh fairly

frequently because I like to go visit

our offices and I like to go visit our

our hugging people um but yeah I I uh

keep my LinkedIn pretty up to date about

what's happening at hugging face and and

what I'm up to next cool uh do you have

any events to be aware of or is that um

good question a couple that are in

Sacramento so those might be a little

too

specific

I I've never been to America so I don't

know that's quite small right is that

like a specific place within California

it is yes so it's the capital of

California but not I'm the only I'm the

only hugging face person here oh W um

and so uh but I'm fairly close to San

Francisco where we have events that

happen fairly frequently um but uh I'd

say nothing nothing for me to plug

until July and when I I'm back in Europe

nice okay I'm going to end the call here

uh and the recording here thank you so

much for speaking to me this morning

this evening whenever yeah I'm gonna

just stop recording yeah okay thank you

so

much yeah it was a pleasure chatting

with you and thanks for doing all this

work that wraps up this week's episode

of unsupervised learning I'm your host

Renee and I've had a great time chatting

with you as always links to everything

we discussed will be in the show notes

make sure you reach out to our guests

questions or feedback reach out to pod

unsupervised learning. until then leave

a like follow or rating on Spotify Apple

podcast or YouTube and until next week

stay curious