Anatole Debierre
Master’s student in Computer Science at EPFL. Previously an ML intern on Uber’s Traffic team. Background in perception and statistics, and a soft spot for mechanism design. I enjoy long puzzles!
Lausanne, CH linkedin.com/in/debierre anatole.debierre@gmail.com
Education
Master of Science, Computer Science
EPFL
- MSc Coursework: Formal Verification, Functional Programming, Machine Learning, Data Analysis
- BSc Coursework: Microengineering (2023 – 2026) — Statistics, Numerical Analysis & Optimization
- Year Abroad: KTH Stockholm, Electrical Engineering and Computer Science (2025 – 2026) — Game Theory
Experience
MLE Intern, Traffic
Uber TechnologiesInternship
- Drove research across a traffic prediction stack ingesting 100k+ GPS pings/sec100k+ GPS pings/sec Floating car data: GPS positions from vehicles already on the road, used as sensors for streets that have none. globally, spanning learned road network representations and inference of unobserved traffic state from floating car data
- Built the team’s backtesting pipeline for evaluating research models against production predictions
- Trained 200+ research models over the internship
- Designed a traffic light phase estimation system reaching ~2.0s transition timestamp MAE~2.0s transition timestamp MAE Mean absolute error of the predicted moment of a light's next phase change, from past traffic and its deduced fixed cycle length, at the limit of what our reference data could measure.
- Interned alongside full-time coursework, and received an invitation to stay full-time
- Extended beyond scope into map matching, implementing ideas from the literature
Research Engineer
UNIL Sports & HealthPart-time
- Developed multi-camera vision system for movement analysis, cutting manual work from 5hrs to 2min
- Applied epipolar geometry to reconstruct joints from noisy data in a way that’s robust to obstructions
- Enabled multi-player tracking with Kalman filters, and built a dynamic dashboard to explore the field in 3D
- Each camera sees joints in 2D; matching them across views recovers a joint one camera can’t see
Teaching Assistant, Probability & Statistics
EPFLPart-time
- Taught ~50 students, using interactive notebooks to strengthen probabilistic reasoning
Awards & Projects
IMC Market-Making Competition
- Placed 1st of >50, thanks to a mix of mildly-quick maths, lots of automation, and investigating and exploiting mistakes in other participants’ algos
- Built my own trading UI in an hour, while trading and writing algos, adding the quick entry and keyboard shortcuts the default one lacked
Jane Street Portal Program
- Selected as one of 30 from >3,000one of 30 from >3,000 An acceptance rate under 1%. for a week-long educational program focused on research
- Gained practice in trading on the farm, specializing in pumpkins and oranges
Apple WWDC Scholar
- Built apps simulating electoral systems and their failure modes; invited to Apple Park
- Arrow’s theorem: no ranked voting rule meets a few fair-sounding conditions at once
- A favourite of mine, approval voting: approve as many candidates as you like, so similar ones stop splitting the vote
Mechanism-design GPU marketplace (CalcX)
- Hackathon project: a live, zero-trust market renting idle consumer GPUs for verifiable computation
- The exhilarating part was designing the incentives: GPUs bid in an open auction, each staking money on its answer
- Answers are cheap to check, so anyone can audit one, and a wrong answer forfeits its stake
- Judges from OpenAI and YC startups preferred pooling trusted machines, like a company’s own laptops, to an open market
Study-abroad assignment predictor (ExSight)
- Reimplemented EPFL’s assignment algorithm so students could see their odds from GPA and preferences
- Drew 700+ sign-ups, more than the 600–650 eligible, becoming the de facto tool for exchange planning
Additional Information
Skills
- Agentic SWE
- Python (NumPy, Pandas, OpenCV)
- C++
- TypeScript
- Scala
- Spark
- SQL
- FPGA
Interests
- Effective altruism
- Mechanism design
- Voting systems
- Rock climbing
- Urban mobility
- Losing at poker





