Design experiments that test how AI responds to legal rules and develop ways to evaluate the results.
The role
Design experiments exploring how AI systems respond to legal rules and competing obligations. You will build evaluation environments, training pipelines and runtime controls, working with legal specialists to ground the research in practical legal situations. The remit includes owning projects through publication and releasing benchmarks or tools that others can use to assess results.
What you will need
At least one year of research or machine learning engineering experience, alongside a relevant degree or equivalent practical experience. Strong Python and modern machine learning skills are required, with evidence of work in AI safety, alignment or evaluation. Legal training or experience assessing model based judges would be useful. The detailed brief requires US work authorisation without sponsorship.
What is on offer
- A full time, office-based San Francisco role, with attendance five days each week.
- Base salary is $185,000 to $250,000, with equity offered separately and responsibility for developing original research.
#J-18808-Ljbffr