Research Engineer AI Safety and Alignment

Skills

About the Role

You will develop safety and alignment evaluations, research alignment, value-learning, and interpretability techniques, and conduct adversarial testing of large language models. You will analyze and mitigate harmful model behavior using approaches such as RLHF and fine-tuning, write production-facing and training code, and help translate research into scalable solutions.

Requirements

• PhD or equivalent experience in Computer Science, Machine Learning, or a related field
• Ability to write clear production-facing and training code
• Experience working with GPUs for training, serving, and debugging
• Experience with data pipelines and data infrastructure
• Strong understanding of transformers and reinforcement learning, including their safety implications

Responsibilities

• Develop evaluation methodologies and metrics for large language model safety and alignment
• Research alignment, value-learning, and interpretability techniques
• Conduct adversarial testing to identify model vulnerabilities and failure modes
• Analyze and mitigate bias, toxicity, and harmful model behavior using RLHF and fine-tuning
• Translate safety research into scalable solutions and best practices
• Contribute to AI safety research through publications and presentations

Benefits

• Equity
• Health coverage for employees and their families with most premiums covered
• 401(k) contribution
• Paid parental leave of up to 20 weeks
• Four weeks of PTO
• Daily in-office catering
• Monthly DoorDash stipend
• Monthly wellness stipend