Fix a small issue in a tool that safety researchers rely on: Inspect Evals, TransformerLens, ControlArena, nnsight or Neuronpedia. No ML research experience needed, just software skills.
Good tooling multiplies everyone else’s research. A merged pull request is also concrete, public proof that you can contribute to the field.
Output: One or more pull requests, ideally merged.
Put two or three 2–4 hour blocks in your calendar. Short, focused sessions beat scattered evenings.
Pick one tool from the links below. Read its README and contributing guide, and get it running locally.
Filter the issue tracker for beginner labels. Comment that you are working on one before you start.
Make the smallest useful change, add a test if the project uses them, and open a draft pull request so maintainers can steer you early.
Address feedback quickly. When it’s merged, write a short post about what you learned.
All steps done
Share what you made, then try another project. The PDF version has room for notes.
Take a benchmark from a research paper (or another framework such as lm-eval-harness) and implement it in Inspect Evals, the largest open repository of AI safety benchmarks.
Pick a safety paper with public code and make it reproducible: fix broken dependencies, write setup docs, or package it so others can run it in one command.
AI control asks: if a model might be trying to sabotage us, can we still use it safely? ControlArena is the UK AISI’s open toolkit of settings, model organisms and protocols. Run one setting, then improve a monitor or document a gap.
Write a small, repeatable test that checks whether a model admits when it does not know something, using Inspect, the open-source evaluation framework used by the UK AI Security Institute.
Use Neuronpedia, a free browser tool, to see which internal “features” of a language model light up for different ideas, then try to break your own explanation of one.
