Post-Training AI Research Engineer

Harvey, Inc.; San Francisco (CA); On-site; USD 231,000 - 340,000; Full time

Job summary

Harvey is seeking a research engineer in San Francisco to scale post-training loops and run training experiments. You will interpret results and collaborate with internal and external research partners to build better data, environments, graders, and training recipes.

The ideal candidate has hands-on experience training open-weight models in research or production settings. You will manage ambiguous applied research projects and communicate clearly with researchers, engineers, product teams,

Qualifications

• Hands-on experience with post-training or model-training work (e.g., SFT, RLHF/RLAIF, reward modeling).
• Strong judgment about model behavior and ability to read traces and identify failure modes.
• Strong Python and research-engineering ability; write clean, debuggable code.

Responsibilities

• Drive post-training experiments, pushing agent performance while balancing cost, latency, security, and governance.
• Optimize agent harnesses, including domain-specific skills, tools, subagents, and retrieval strategies.
• Design and develop grading and reward systems for reliable evaluation in high-stakes tasks.
• Study agent behavior and translate findings into training data, evals, or harness changes.
• Collaborate with Harvey researchers and external partners to define experiments and move toward concrete model improvements.

Skills

Post-training experiments

Model training

Python

Experiment design

Communication

Tools

SFT

RLHF/RLAIF

Distillation

Job description

Harvey is seeking a research engineer in San Francisco to scale post-training loops and run training experiments. You will interpret results and collaborate with internal and external research partners to build better data, environments, graders, and training recipes.

The ideal candidate has hands-on experience training open-weight models in research or production settings. You will manage ambiguous applied research projects and communicate clearly with researchers, engineers, product teams,