AI Alignment Research Engineer — Evaluation & Safety

W3 Sourcing; San Francisco (CA); Hybrid; USD 140,000 - 210,000; Full time

Job summary

W3 Sourcing is seeking a Research Engineer for AI Alignment & Evaluation in San Francisco, CA. You will independently design and build evaluation environments for frontier AI models and own the full lifecycle from concept to refinement.

You will work with advanced LLM agents, review outputs for subtle errors, and exercise strong judgment to ensure safe, reliable behavior. This hybrid in-person role offers substantial ownership and collaboration with a tight technical team.

Qualifications

• 1+ years of software/ML engineering or closely related work.
• Proficiency with Python.
• Strong interest in AI alignment, safety, or security.

Responsibilities

• Design complex evaluation environments for frontier AI models.
• Own evaluation projects end to end from ideation to refinement.
• Investigate potential model failure modes and how agents may exploit constraints.
• Develop and improve software infrastructure to isolate, reproduce, and evaluate behavior.
• Work with LLM-based agents to accelerate research workflows.
• Review agent-generated outputs to identify subtle errors.
• Build long-horizon tasks near current model capabilities.
• Apply qualitative judgment when metrics miss nuanced behavior.
• Learn unfamiliar domains quickly as required by environments.
• Share findings and context with the research and engineering team.

Skills

Python

Software Eng

ML Eng

AI Alignment

Job description

W3 Sourcing is seeking a Research Engineer for AI Alignment & Evaluation in San Francisco, CA. You will independently design and build evaluation environments for frontier AI models and own the full lifecycle from concept to refinement.

You will work with advanced LLM agents, review outputs for subtle errors, and exercise strong judgment to ensure safe, reliable behavior. This hybrid in-person role offers substantial ownership and collaboration with a tight technical team.