Ibragim Badertdinov
I build evals and RL envs for SWE and research. Research Engineer at Nebius, London. - Lead author of SWE-rebench — NeurIPS 2025, 12M+ HF downloads - SWE-rebench leaderboard — 1M+ visits per month - SWE-rebench V2 — 32,000+ RL environments in 20 languages, ICML 2026 - Speaker at AI Engineer Europe 2026 Dentistry → Healthcare Mgmt → ML/NLP → Research / Open source / RL envs Now Comparing coding agents on my everyday tasks → leaderboard Ask me about SWE benchmarks, RL environments, agent evals Open to Research collaborations, talks, help with SWE / RL envs ---
PROJECTS
SWE-rebench Leaderboard – Live benchmark for coding agents on fresh GitHub tasks, refreshed monthly. SWE-rebench – 21,000+ executable Python SWE tasks for RL training of code agents. SWE-rebench V2 – 32,000+ executable SWE tasks in 20 languages + 126,000+ PR-based tasks. Project details → ---
SELECTED PAPERS
Benchmarks SWE-rebench (NeurIPS 2025), SWE-rebench V2 (ICML 2026) RL training Long-context multi-turn SWE agents (2025) Search Guided search for SWE agents (ICML 2025) Data Scaling data collection for SWE agents (2024) All selected papers → ---