Research Engineer, Infrastructure

The role

You will build systems to adapt and improve open-weight models for legal work. This includes owning infrastructure for distributed training, RL rollout systems, and internal tools for data management and training job tracking.

What you will do

• Scale the post-training stack supporting SFT and RL on open-weight models.
• Build distributed rollout infrastructure for model inference and tool execution.
• Build infrastructure for reproducible agent environments.
• Own performance and reliability of training runs including GPU utilization and failure recovery.
• Identify and remove bottlenecks across training and inference.
• Build internal tools for training data management, evals, and tracking training jobs.
• Integrate models into production including serving, latency, and cost.
• Partner with researchers to turn experimental workflows into dependable systems.

Required qualifications

• Experience building ML infrastructure, distributed systems, or research tooling.
• Ownership of performance, reliability, and operation.
• Strong Python and generalist software engineering skills.
• Familiarity with PyTorch, JAX, or similar ML frameworks.
• Experience with distributed training or inference.
• Ability to diagnose performance bottlenecks and failures across the stack.
• Practical judgment about what to build versus adopt.
• Ability to partner with researchers and ship quickly.

Nice to have

• Experience with post-training workflows
• Experience with RL rollouts
• Experience with agent environments
• Experience with sandboxed tool execution

About the company

Harvey is transforming how legal and professional services operate by combining frontier agentic AI, an enterprise-grade platform, and deep domain expertise.