Senior Research Scientist, Post-training and Safety
Compassion Aligned Machine Learning (CaML)Remote (US)
AI safety & policy
Full-time
AI safety & policy
Full-time
Salary
$200,000 to $300,000
Routes to impact
Direct high impact on an important cause
Description
Own the training side of research on how values are instilled in LLMs and whether they survive post-training.
- Run RL post-training on open-weight models, including RLHF, RLAIF, DPO, GRPO, and RLVR
- Lead mid-training and continued pretraining using synthetic document generation, controls, and mixing-ratio experiments
- Build evals that test whether model reasoning holds up under adversarial prompting and distribution shift
- Study value degradation through training, including reward hacking, sycophancy, specification gaming, and generalisation
- Write up research for arXiv and LessWrong
This text was generated by AI. If you notice any inconsistencies, please let us know using this form.
