Notes on building better AI data
Practical, research-backed articles on evaluation, training data and the human work behind both. Every claim links to its source.
Egocentric data collection for robotics and embodied AI
Why first-person video matters for models that act in the world, what the major public datasets teach about scale and annotation, and what a capture program needs to get right.
EvaluationWhat makes an AI benchmark trustworthy
Contamination, unfair tests and vague task statements quietly break evaluations. Here is how to build benchmarks whose scores you can believe.
Training dataRLHF and preference data, explained for teams that collect it
Reinforcement learning from human feedback is only as good as the comparisons behind it. What to collect, who should collect it, and how DPO changes the pipeline.
Agent environmentsHow to build agent environments that test real work
Docker images, terminal tasks and verifiers: a practical anatomy of the environments used to train and evaluate AI agents.
Training dataQuality over quantity in fine-tuning data
The research case for smaller, cleaner, expert-written datasets, and a practical checklist for getting there.
Data collectionExpert or crowd? Measuring quality in human data collection
When crowd labels are enough, when you need specialists, and how to measure annotation quality instead of assuming it.
SafetyRed teaming language models: methods and lessons
What the research says about finding failures before users do, and how to run adversarial testing that produces usable training data.