Founded nearly a decade ago in Paris, Hugging Face has quickly become a major French success story in a buzzy field: open source AI.
The company, which has built a platform that hosts AI models and provides tools for engineers to build, train and deploy the technology, has established itself as a major player in the global open source community.
With nearly $400m raised to date from high-profile VCs like Sequoia Capital, as well as some of tech’s biggest names including Google, Amazon, Nvidia and Intel, the French scaleup has successfully built up a huge presence in the US, where it opened an HQ immediately after launching. Hugging Face now has over 200 employees across the world, says talent acquisition lead Flavien Coronini, of which 80-90% are in technical roles like machine learning engineers and machine learning researchers.
These are coveted jobs. “As an open source platform, Hugging Face is community-driven,” says Coronini, “so we have great visibility externally. […] This means we get a lot of inbound applications.”
Sifted sat down with Coronini to find out what the company looks for in future employees. This conversation has been edited for clarity and length.
What qualifications do you look for in candidates?
Educational background isn’t a huge performance indicator for us. We recognise that AI experience can be gained through a variety of means, including formal education, self-study and practical experience.
What I usually look at is what people do on the side: open source projects, contributions to different open source communities, technical skills. Anything showing they are able to think outside of the box.
You’ve built a huge community of open source developers. How much do you tap this resource for recruitment ?
We get huge amounts of inbound applications, but a lot of profiles also get to us through our own networks. We are very active in the open source space, where by definition the door is open to anybody to contribute, and we regularly have external people collaborating with us on projects.
It can even be non-technical people helping with more functional things, like technical documentation. This means we can literally see their impact on the project, it enables us to evaluate their performance and we often recruit just like that. I’d say this represents about 30-40% of all our hiring, and it means we’re constantly recruiting.
I’m often asked what’s the best way to join Hugging Face. I always recommend being visible in the open source community, by contributing to pull requests (discussions about proposed changes to code) and collaborating on projects with our engineers. It’s the best way to show your added value and impact.
Does this mean candidates with no open source experience stand no chance?
It’s better to have engaged with the open source community, but we don’t rule out those who haven’t because there are unfortunately many closed source companies who do not authorise their employees to work on open source projects.
If this is the case, I’ll be looking at candidates’ professional experience, what projects they worked on, what applications they’ve built. Sometimes people also have open source projects held privately, which can’t be seen by the wider community, so we can dig into that too.
What kind of professional experience do you look for?
Professional experience remains a key criteria, not just for what the candidate has built, but also to assess if they have worked in environments that can be transposed to Hugging Face. We have a strong culture that is not for everyone: we are very lean, very flat, there is little hierarchy and no bureaucracy. We don’t have many structures and processes. We try to do the minimum for the company to function and scale, but we don’t take ourselves too seriously. So we need people who can adapt to that.
In a similar vein, we work asynchronously because we are remote-first. For a given project, you’ll have a team of people spread across the world but working together daily. We communicate mostly in written form and we do very few meetings. It means teams are really autonomous and have a lot of leadership, so we need someone who aligns with this.
This is why we prioritise people who have worked in startups and scaleups — because Big Tech is a wholly different world. If someone working at Google wants to work at Hugging Face, I’ll be asking myself if they truly realise where they are headed, and that we have none of the structured processes they currently have.
What’s the recruitment process like?
Recruitment at Hugging Face is decentralised, meaning anyone at the company can source profiles and everyone is encouraged to headhunt and recruit. Sometimes, a team will start a recruitment process without even keeping me in the loop, and I only intervene in the final stages to evaluate culture fit or determine the salary. We do this because we know our employees and their networks are the best way to source candidates, while also being the best advocates for the company.
It means the recruitment process changes from one candidate to another, but we generally have three stages. There is an assessment of culture fit, usually the first stage, where we screen the profile to ensure they share our vision and are excited by Hugging Face. Then we will have a technical assessment — usually a take-home assessment based on a typical use case they could face as an engineer at Hugging Face.
For a front-end engineer, for instance, we could ask them to look at a page on our site and suggest improvements. We’ll be looking at how they analysed the problem, how they would have resolved it and their ability to drive things forward and make proposals. We’ll also be assessing how they communicate asynchronously, in writing.
The last stage is usually an interview with team leads or the cofounders.
What are the first few weeks and months like at Hugging Face for new recruits?
We don’t have a pre-defined model or structured process for induction. Employees are free to function as they wish.
The team lead will usually integrate the new recruit and define priorities, and we also have a buddy system to match the employee with someone else at the company. But almost from day one, we like to have people who will tell us where to go, rather than people we need to give directions to. Employees must have a lot of independence and autonomy.
Read the orginal article: https://sifted.eu/articles/hugging-face-hire-how-to/
