Job Details

About the Company

An early-stage AI research lab focused on interpretability, alignment, and reinforcement learning is hiring a Research Engineer. Founded by former frontier-model researchers, the team works directly on model internals and training dynamics to better understand how AI systems reason. The lab runs fast experimental research cycles, building custom tools to explore open-ended questions about model behavior.

About the Role

This role focuses on building the experimental tooling that enables interpretability research. You will develop systems that allow researchers to inspect, measure, and manipulate internal model representations. This is not a production ML or MLOps role — it's for engineers who enjoy building new experimental systems from scratch and working closely with researchers.

Responsibilities

• Build custom RL-style environments and experimental testbeds for interpretability research
• Develop tools for activation tracing and mechanistic interpretability
• Implement probes to detect latent concepts (e.g., deception, goals, uncertainty)
• Prototype activation-level steering methods beyond prompting or fine-tuning
• Collaborate with researchers to rapidly iterate on experiments and tooling

Qualifications

• Strong programming skills in Python
• Experience with ML frameworks such as PyTorch or JAX
• Experience working with deep learning models or LLMs
• Strong software engineering fundamentals and ability to build experimental ML systems

Preferred Skills

• Experience in mechanistic interpretability or alignment research
• Familiarity with reinforcement learning environments
• Background working with model internals, representations, or probing methods
• Research experience (PhD preferred but not required)

Pay range and compensation package

Competitive salary, equity, and benefits.