OpenAI
Researcher, Safety Training
- London
- Permanent
Conducts research on safety post-training methods such as RLHF, RLAIF, DPO, PPO, reward modeling, adversarial training and synthetic data, then builds evaluations for harms like jailbreaks, hallucination and policy adherence. Works cross-functionally with core training, product, policy, legal and trust and safety teams to ship safety improvements into production models.
OpenAI