An unsolicited guide to good research

Eugene Vinitsky · March 2026

A grab bag of ways to be a better researcher

Research Advice

How to think about human-likeness in the age of autonomy

Daphne Cornelisse · February 2026

A critical examination of how we measure 'human-like' behavior in autonomous driving benchmarks, arguing that current metrics incentivize reproducing human mistakes rather than achieving safe, realistic driving.

Autonomous DrivingBenchmarks

How to catch subtle RL bugs before they catch you

Daphne Cornelisse · October 2025

Practical strategies for identifying silent errors in reinforcement learning codebases, bugs that don't crash but produce incorrect results, illustrated through code reviews, testing, visualization, and performance benchmarking.

Reinforcement LearningEngineering