Arthur Bigot
MSc student in Data Science at ETH Zürich, currently a Research Fellow at Harvard University’s Shakhnovich Lab and a Summer Research Fellow at CBAI. I study models for language, vision, and proteins, focusing on AI safety, alignment, and mechanistic interpretability.
Publication
Viral Proteins Reveal Geometry of Protein Language Models. Accepted at the ICML 2026 GenBio Workshop and ICML 2026 FM4LS Workshop. arXiv:2606.12609.
Research
- Sparse Autoencoder Latents for Concept-Based Art Retrieval
- Mechanistic Interpretability on SafetyPolytope
- G1bbon Benchmark
- Real-Time Vehicle Tracking for Autonomous Racing