Lille, France
Mikhail Doroshenko
AI researcher & ML engineer
I work on frontier-model evaluation, technical forecasting, computer vision, and tool-augmented agents.
Work
- Empirical evaluation and testing of frontier AI models.
- Technical forecasting of AI progress.
- Computer vision, recommendation systems, and agents.
- Applied optimization, including C++.
Selected work
-
Co-author; contributed 50+ expert-level questions to the benchmark.
-
Unified LLM rankings from public benchmarks — 1,700+ models across 5,000+ benchmark tables, updated daily.
Experience
-
Machine Learning Engineer — MCQ-scan
Tool-augmented agents, production computer vision, and recommendation systems.
-
Research Engineer — Navigine
High-performance C++ indoor-navigation library; particle filters, Kalman filters, sensor fusion.
Besides work
- Top 3 at a Blaston VR tournament (VR MEET Europe 2024).
- Forecast on Metaculus and Manifold.
- Table tennis, StarCraft, and calisthenics on the side.