About

Burgundy is not only the primary color of this website, it is also my birthplace and where I essentially spent the happy first 18 years of my life within a wonderful family. The irony being that this colour is named "Bordeaux" in French, which is a different charming wine region of France. Or so I've heard. I couldn't possibly know, I've never been there.

During my teenage years, I knew I liked two things: math and video games. I decided to study math and went to the CPGE Lycée Louis-Le-Grand in Paris. CPGEs are like the preparatory program to the French equivalent of the Chinese Gaokao, i.e. a big exam that determines which university you can attend; but contrary to the Chinese counterpart, the competitive exams are taken on university-level material, after two years of post-high-school study. This was a memorable period, and I believe this program fundamentally shapes us because it teaches us how to repeatedly handle failure. On top of that, the friends I've made there are still among the best I have.

After the exam, I landed in École polytechnique. It's a special kind of university: during the first month there, I was sent into the middle of the Empty Diagonal to learn the basics of military training, and then was sent to the Gendarmerie Nationale in Colmar drinking Glühwein on the Christmas market. I didn't understand exactly what the military experience would bring me at the time, but as I get older, I get back to it from time to time. It made me understand more deeply what "comfort zone" means. "Where there is a will, there is a way" was the motto of our Year Group Commander, and as cliché as it may sound to the know-it-all readers (not you, the other ones), I felt it in my veins when I had to push for SIGGRAPH deadlines.

Back on campus, I decided to go for computer science. It was like math, but useful with a faster feedback loop. I had the chance to intern in Singapore for 12 weeks, and also accompanied a team to the Microsoft Imagine Cup with Theatrall, a project that had the goal of providing real-time subtitles on smart glasses during a theater play. I had no clue about machine learning at the time, so I went with the good old Levenshtein distance algorithm to match what was detected to the text of the theater piece. I wasn't really an entrepreneur, but one of my teammates clearly had the itch. I didn't exactly understand him at the time, but I felt a mix of fascination and respect for his audacity.

I studied theoretical computer science at École polytechnique. The more I dug into it, the more I found it somehow... dry. SAT solvers are interesting to an extent, but it wasn't enough of a dopamine hit. Fortunately, we are encouraged to do our last year in another university, so I went for a master's at ETH Zurich, which was an opportunity to explore a different field. I was naturally drawn to computer graphics, and during my studies, I fell in love with it. You can do any kind of math within this field, and you can hone your programming skills without limit. It's beautiful. It's interactive. It's useful. Que demande le peuple?

I emailed Olga Sorkine-Hornung to ask about doing a master's thesis with her because I liked the topic of her lectures the most (shape modeling and geometry processing), and had the chance to work on how to create a developable approximation of any given mesh. When I realized that I could actually get paid for working on such cool problems, I applied for a Ph.D. there. This was one of the best decisions I ever made, because the four and a half years I spent at IGL allowed me to find what I wanted to do with my life.

From the beginning of my Ph.D., I oriented myself toward modeling systems. How can people create full-fledged animations, video games, or movies? What are the tools that they use to do so? And as a scientist, what contribution can I make to help people be more creative? I started with classic geometry and developed an algorithmic method delivered through a mini-CAD software to create curves that respected specific desirable design properties. I realised that data-driven techniques could be immensely powerful, so I researched how I could turn sketches into 3D objects. This sounded like a dream for modeling systems: take something that's easy to do (a rough sketch), and turn it automatically into something that's hard to get (a 3D model). But what caught my interest even more was the specific problem of the shape representation that can be used for creating shapes, which led me to work on TetWeave. It's also the hardest project to explain. In short: it's a point cloud that can be turned into a mesh, but it's better for gradient-based optimization.

The relevance of geometry is challenged by generative AI techniques that operate in a purely rasterized format. To put it bluntly, some argue that geometry might have become a fake task for digital creations. I don't know if that's true, but it is almost surely not if geometry is the target itself rather than an intermediary step, which is the core of digital fabrication. This leads to works like SD-πXL or ALPS that leverage a pre-trained network for optimizing the parameters of a specific representation. The representation is built from geometric primitives within a constrained color palette. In short, we use the semantic understanding of AI models to optimize, rather than relying on geometric approximation alone.

My interest in this constrained generation led me to the field of embroidery design. Geometric primitives can be mapped to stitches, and the color-palette constraint maps directly onto embroidery's natural limitations (you have a limited amount of thread, by definition). Embroidery is beautiful and artistic, yet particularly complex for many people to create.

Developing tools that help others be more creative has slowly but surely become my life goal, and I find myself more fulfilled by working on CAD software that enhances people's creative power. I decided to venture into this direction and create Boldbobbin, a modern platform for embroidery design.

Outside of work, I am learning Mandarin Chinese so that I can finally exchange a few sentences with my wife's family. I've also been doing some calisthenics for four years. I'm not particularly impressive at either of these things, but I have been practicing them very consistently, which I've come to think compounds over time.

And that's basically it! Thank you for reading, I will update this page as my life unfolds and write more specific content in the blog section of this website.

I wish you, dear reader, a wonderful day.