OpenAI
Researcher, Safety Training
$380K–$500K
San Francisco +1 · Full-time · Posted May 25, 2023
- Location
- San Francisco
London, UK - Education
- PhD in machine learning, computer science, or a related field, or equivalent practical research or engineering experience.
About the job
Research and implement post-training methods, evaluations, and product interventions that improve the safety, reliability, and alignment of OpenAI models.
What you'll do
- Research and implement safety post-training methods and model training interventions.
- Build evaluations and measurement infrastructure; analyze model failures and safety risks.
- Collaborate across research, product, policy, legal, and safety teams to incorporate improvements into training and product launches.
What they ask for
- Hands-on experience training, fine-tuning, or post-training language models, multimodal models, or agentic systems.
- Experience with reinforcement learning or post-training methods such as RLHF, RLAIF, DPO, PPO, reward modeling, preference optimization, or instruction tuning.
- Strong research judgment and engineering ability, including running experiments and working with large-scale ML systems.
- PhD in machine learning, computer science, or a related field, or equivalent practical research or engineering experience.
Nice to have
- Experience improving production model behavior through post-training or deployment feedback.
- Experience building safety evaluations, automated graders, red-teaming systems, or monitoring pipelines.
Pay
$380K – $500K USD base salary
Tools
RLHF, RLAIF, DPO, PPO
AI focus
Develop safety post-training methods, training interventions, synthetic data, and evaluations to improve model behavior and alignment.
Model-written summary · the original posting is authoritative