Research Scientist, Frontier Alignment & Post-Training

Role Overview & Mission

OpenAI is seeking an experienced Research Scientist to join our Frontier Alignment and Post-Training team in San Francisco. In this role, you will lead research and algorithmic improvements for next-generation frontier foundation models, ensuring they remain reliable, steerable, robustly safe, and aligned with human intent across complex multi-step reasoning tasks.

Key Responsibilities

• Design, implement, and scale novel reinforcement learning from human and AI feedback (RLHF/RLAIF) algorithms for reasoning models.
• Investigate generalization phenomena, self-correction dynamics, and scalable oversight techniques across multimodal domains.
• Collaborate closely with distributed systems engineers to execute large-scale distributed training runs across high-density GPU clusters.
• Publish foundational findings, open benchmarks, and technical safety reports contributing to safe superintelligence development.
• Evaluate empirical alignment failures, red-teaming vectors, and unintended behaviors in pre-deployment frontier model checkpoints.

Qualifications & Experience

• PhD in Computer Science, Machine Learning, Applied Mathematics, or equivalent practical research track record (4+ years post-undergrad ML research).
• Demonstrated track record of first-author publications at top-tier venues (NeurIPS, ICML, ICLR, ACL) or comparable contributions to production foundation models.
• Deep expertise with PyTorch, distributed training frameworks (Megatron, DeepSpeed, Ray), and CUDA kernel profiling.
• Strong theoretical grounding in reinforcement learning, probabilistic modeling, and optimization.

Preferred Skills

• Hands-on experience training models larger than 70B parameters or fine-tuning frontier reasoning systems.
• Experience developing synthetic data curricula and automated verifiers for mathematical and algorithmic problems.

Compensation, Benefits & Equity

• Base Salary: $245,000 – $385,000 per year, based on experience and level.
• Substantial Participation Units (equity-equivalent) in OpenAI with annual refreshers.
• Comprehensive medical, dental, and vision insurance with 100% employer-covered premiums for employees and dependents.
• Unlimited paid time off (PTO) and company-wide seasonal refresh breaks.
• Generous 401(k) matching, fertility benefits, and continuous executive education stipends.