Machine Learning for Alignment Bootcamp (MLAB)
The Machine Learning for Alignment Bootcamp (MLAB) is a program of Redwood Research designed to expand the pool of ML-proficient AI safety researchers. Held in Berkeley, CA, MLAB runs for approximately three to four weeks and combines pair programming, lectures, and hands-on curriculum to bring participants from strong programming backgrounds up to speed on modern deep learning, transformer architectures, and interpretability techniques. Two cohorts ran in 2022 (January and August-September), with a combined ~68 participants, and several alumni went on to full-time alignment research roles. The curriculum materials are publicly available on GitHub and have been adapted by other organizations worldwide.
Funding Details
- Annual Budget
- -
- Monthly Burn Rate
- -
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
- Fiscal Sponsor
- -
Theory of Change
MLAB operates on the theory that a key bottleneck in AI alignment research is the shortage of researchers who combine alignment motivation with the practical ML engineering skills needed to work on modern large language models. By running intensive, highly selective bootcamps that give technically strong programmers hands-on experience building and analyzing transformer models, MLAB creates a pipeline of researchers capable of contributing to mechanistic interpretability, adversarial training, and other empirical alignment approaches. The causal chain is: identify motivated people who lack ML depth, rapidly upskill them via immersive peer learning at the Constellation alignment hub, connect them with Redwood and peer organizations during the program, and thereby accelerate their entry into full-time alignment research. Publishing the curriculum openly multiplies impact by enabling other groups to run similar programs.
Grants Received
No grants recorded.
Projects
No linked projects.
People
No linked people.
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.
Details
- Last Updated
- Apr 2, 2026, 10:09 PM UTC
- Created
- Mar 19, 2026, 10:31 PM UTC