About me
I'm Guillaume, a Machine Learning Engineer based in Paris 🥐, currently exploring the startup world. I have interests in both Language Models and Computer Vision — I might pick one some day.
I spent nearly three years at Google Zürich, developing Machine Learning models for various features of Google Search.
Before that, I completed dual master's degrees at École Polytechnique 🇫🇷 and ETH Zürich 🇨🇭, where I focused on machine learning, algorithms and data science.
Always open to new challenges — whether it's building something ambitious or just something fun. Feel free to reach out!
My latest contributions
See all →- Metropoly: a game to test your knowledge of the Parisian metro
To help me memorize all the metro stations in Paris, I made this small game to test my knowledge. More features got added along the way, as people around me were really enjoying playing. Just try it out !
ReactAWSDatabase - Towards Online Real-Time Video Inpainting
State-of-the-art video inpainting models gave great results but were too slow for real time. I introduced a memory mechanism that stores and reuses features across frames, doubling the throughput of three different transformer models. Published at NTIRE 2024 (CVPR workshop).
PyTorchComputer VisionTransformers