Based in Paris, Arnaud Fournier leads OpenAI’s Forward Deployed Engineering team. Created last January, the unit supports OpenAI partners in the co-development of new artificial intelligence solutions.
JDN. You launched OpenAI’s Forward Deployed Engineers team from Paris this year. What is its primary mission?
Arnaud Fournier. We created this team last January, so almost ten months ago now. I had just returned from New York at the time. The objective of this team is to work with our strategic customers, or rather our partners, in cases where OpenAI aims to create new products or services, or to establish research partnerships with companies. It is a team that has a global presence at OpenAI: we have collaborators all over the world, and particularly in France. With around ten people, Paris is today the largest office in the world for the FDE team at OpenAI. It was a logical location for us, both for the quality of talent and to cover the European area. This team has continued to grow over the past ten months and will continue to develop in the coming months and years.
What types of projects are you actually developing within the FDE team, and with what objectives?
We realize that we are at the beginning of the history of artificial intelligence, and companies need support to understand how to use it and what they really need. The vocation of this team is based on two axes.
First, learn as close as possible to our customers and partners. This is not a model that we sell as a service, it is truly a partnership between OpenAI and the company, in the form of co-development. This can be research or co-development of solutions. There is therefore a first learning challenge, from which the development of new solutions and new offers arises. The Forward Deployed Engineers team aims to create the first versions of these products, which can tomorrow be integrated into our OpenAI offers for the general public, businesses or developers. To cite a few examples, we worked on the first version of what became the Agent SDK, the tooling for creating OpenAI agents, which is now public. Our team created the first version, which we then open-sourced as Swarm, before developing it as an Agent SDK with our product teams, who took the project over to make it a traditional OpenAI product.
“We worked on the first version of what became the Agent SDK”
Then, we carried out projects that were close to OpenAI’s heart within this partnership framework. We can cite for example John Deere, on which I personally worked during the first months of creating the team. The objective was to see how we could support John Deere, a company which notably develops tractors, to help farmers around the world reduce their use of chemicals, a major societal issue. Using artificial intelligence, we advise farmers on how to better deploy their weedkillers and use just the right amount. This has a significant impact, both for society and for the farmers themselves. This is also part of OpenAI’s mission.
Do you only support large groups or do you also work with start-ups?
We work with companies that want to innovate and push the capabilities of what exists today. These are not just large groups, but we actually find many large groups and innovative start-ups. This covers a wide variety of industry sectors: health, energy, manufacturing, semiconductors… In short, we touch on a spectrum of important subjects which always come down to important issues for OpenAI, in the sense that they serve our mission. So we work with these companies and are open to partnerships, but there really has to be this dimension of innovation and willingness to push existing capabilities.
We observe a certain French tropism, both within the FDE team and more broadly at OpenAI. How do you explain this strong presence of French talent in the company?
What is quite interesting is that we have a lot of French people at OpenAI, in all organizations and at all seniority levels. There is a marked French tropism, particularly in the FDE team. We have them in the United States and France. For this team, our recruitment logic is mainly regional. We have collaborators across North America, EMEA and Asia Pacific. In these regions, we recruit where the talent is and we have the ability to be flexible on their location. It turns out that Paris, due to the quality of the engineers and our installation in France, was one of the most logical places. We also have collaborators elsewhere, for example I have a team in Munich, and we work across the entire EMEA region. But Paris was a place where we managed to recruit very good quality people, and it was interesting for OpenAI’s place in France.
Does the FDE team continue to strengthen? Are you still in the recruitment phase in Paris?
At the Paris office, we continue to actively recruit until the end of the year. We have many positions open. If we look at the Paris office as a whole, it is mainly an engineering office. We of course have more traditional profiles, on the business side, but the main thing is engineering roles that cover different functions. We have FDEs, but also other engineers who support, for example, start-ups to use our technologies, engineers deployed for our enterprise clients, customer success teams, etc.
