Build track
You build the environments: the tasks, the scoring, and the systems that train and test a model. CS, EECS, Math, Stats, Data Science, IEOR, or any major where you learned to program well.
Reinforcement learning at Berkeley
Research
The strongest technical students on campus, training agents on real hardware.
Engineering
Experts from every major at Berkeley, labeling data for you.
Tracks
Build track for technical majors, domain track for everyone else. You don't need to be CS.
Domain track
You write the questions, the answer key and the rubric in your field, then grade what the model produces. No programming.
What we do
- 01
Research, robotics first
A real robot arm trained in simulation, RL for language models, and a weekly reading group.
- 02
Environments and expert data
We build RL environments and expert-labeled datasets for AI labs.
- 03
The Applied RL DeCal
A semester course from the fundamentals to trained agents. Every module ends in a working project.
- 04
Community
Speakers, lab visits, a hackathon, a retreat.
Departments
Research does applied RL. Engineering builds environments and datasets for labs.
Research
Applied RL from simulation to a real robot on the bench. A reading group, research pods, papers.
Engineering
RL environments and expert-labeled data for the labs training frontier models, in the formats they already use.
Majors
Each major labels the data only it can judge. Tap a major to see what.
Philosophy
Domain track
Labels
Grades whether an argument is valid, flags fallacies, writes the rubric for ethics cases.
Example task
Rebut this argument. Graded on logic, not tone.
Applications and projects
Students: any major, any year. Companies and labs: tell us what you need built.