Since raising a record-setting $220m seed round over a year ago, it’s been a wild ride for Paris-based startup H Company.

The startup was created by four ex-Deepmind scientists and computational mathematics student Charles Kantor to build AI agents. One of the hottest companies out there at the time of launch, it quickly hit headwinds — a year later, only one of the original five cofounders remained operational.

Kantor was replaced last June by former Palantir exec Gautier Cloix, who came in promising to focus his efforts on bringing H’s technology into enterprises. To do so, he said, he would follow Palantir’s example and recruit teams of ‘forward deployed engineers’ (FDEs) — software engineers that embed directly with customers to deploy the technology on the ground.

Head of people Audrey Collot and talent lead Geoffroy Hussenot tell Sifted FDEs are the fastest-growing job category at H right now. The company has just hired a head of FDE and plans to have a team of at least 15 people by the start of 2026 (H has a total workforce of 75 employees).

Sifted sat down with Collot and Hussenot to find out what they look for in future FDEs. The conversation has been edited for clarity and length.

Are there specific academic qualifications you look for in FDEs?

Hussenot: Academically we’ll be looking at grandes écoles (top-level higher education institutions in France) like École Polytechnique, as well as prestigious international universities like Stanford, MIT, EPFL and Berkeley. Qualifications are typically masters in data science and computer science, and even quantum physics. We don’t necessarily look for a PhD — that’s more of a pre-requisite for our research team.

Beyond a diploma, we need candidates with proven experience with one or, ideally, several clients. FDEs must have communication and interpersonal skills. Typically, candidates will have led consulting projects with clients where they implemented tools and supported software integrations, with a demonstrated impact. These will be quite technical projects related to data science, with enterprise clients in various industries — pharmaceuticals, banking, insurance, automobile — that have very complex problems to solve.

We are not hiring junior profiles at the moment but people who have professional experience with clients, and who will enable us to maximise the return on investment (ROI) with our customers.

What kind of companies do these candidates typically come from?

Hussenot: All the big consulting firms are interesting and can be a pool of talent. But for this role they need to also have a practical and technical edge. Consultants who have purely worked on strategy, who have done lots of slides and PowerPoints, are not going to fit with the role of FDE. They need to have a lot of technical knowledge, because they’re going to be deploying our solutions with clients that have complex architectures and systems.

We also really like former entrepreneurs who founded a tech company and tested themselves, even if it didn’t quite work out. FDEs need to be entrepreneurial, they need to know how to take risks.

**We hear H has managed to attract a few Palantir employees too… **

Collot: There are some but not only. The idea is not to copy and paste what Palantir does. It’s more about bringing something new to the company.

What’s the recruitment process like?

Hussenot: It’s a pretty typical recruitment process with five rounds of interviews. There is a first interview with me, which is more focused on motivations for joining and pre-qualifying the candidate based on their experience. After that we’ll have two technical rounds. For the first one, we give the candidate a real-life coding problem they need to resolve, for which they can rely on any external tool they like. After that, we’ll do an in-depth interview to test their general knowledge on machine learning. For this one, the candidate isn’t allowed to use any other tools, it happens either on-site or via video call.

The fourth round is with a team lead, who will ask more questions about cultural fit. And the last interview, also largely about cultural fit, is with a member of the executive leadership, like the CTO or CEO.

Our objective is to complete these five rounds in a few weeks, but if we’re looking at a top candidate we’re able to make an offer in just a few days. It’s so competitive at the moment, you have to be fast.

It’s been a turbulent few months for H. Do candidates worry about whether this affects the company’s roadmap and culture?

Collot: There can be questions and transparency is key. It’s true there’s been a phase where we were building the leadership team, but now things are clearer. We recently did an off-site where we clarified our vision, the direction we’re taking, how we want to get there, and the rules and values of the company. This internal work is visible and I think it is distilling externally as well.

Hussenot: Employees who are with us now are motivated and a good sign of that is we are seeing a growing number of referrals, which shows they are acting as ambassadors for the company.

Are there any books, podcasts or courses you would recommend to candidates ahead of applying?

Collot: For this question I consulted (CEO) Gautier Cloix and (CTO) Laurent Sifre. Laurent and the majority of our tech team agreed on a book called Reinforcement learning by Richard Sutton and Andrew Barto. It’s a classic in the field, which helps understand how a machine can make decisions on its own. We include it in our onboarding packs, so it’s a bit of a must-have.

The second book, which we also include in our onboarding packs, is The myth of the machine by Lewis Mumford. That’s Gautier’s recommendation and it is more of a philosophical essay about how we want to make machines more powerful but also make humans more free.

What’s the onboarding process like?

Hussenot: For FDEs specifically, the idea is they are deployed as soon as possible. There’s a bit of learning how to handle our tools but then they are assigned a client very quickly.

Collot: A new recruit would also probably participate in a hackathon within a few months. We really want to accelerate on this. We’ve done a great hackathon already, which generated brilliant ideas — we’re actually beta testing an AI agent for HR functions within my team right now.

More generally, we place a lot of value on new joiners showing curiosity and questioning themselves. We give feedback on a pretty constant basis. You can’t progress thanks to a performance review twice a year, so giving feedback frequently is a key part of how we do things.