Research Engineer – Interpretability Systems (AI)

Location: San Francisco, CA — onsite

Salary: $150,000–$190,000 annually

Company

Early-stage, revenue-generating AI research lab working on interpretability, alignment, and reinforcement learning.

What you will do

• Build experimental systems and custom tooling for interpretability research.
• Perform activation tracing and mechanistic analysis of large language models.
• Develop custom reinforcement-learning-style environments for alignment research.
• Probe internal model representations and detect latent concepts such as deception, goals, uncertainty, and hidden objectives.
• Develop activation-level steering methods beyond prompting and fine-tuning.
• Create benchmarks for model consistency and robustness.

Requirements

• Strong software engineering fundamentals.
• Experience with experimental machine learning or research systems.
• Comfort working close to model internals.
• Interest in interpretability, alignment, reinforcement learning, or mechanistic understanding.
• Ability to work on ambiguous problems and build new tools from first principles.

Nice to have

• PhD in a relevant field.

Culture & Benefits

• Fast, experimental, greenfield research environment.
• Short research cycles focused on building tools, testing ideas, and evaluating results.
• Focus on research systems rather than production ML, MLOps, or large-scale training infrastructure.