4 hours ago
Base Salary
$230k - $340k/yr
Responsibilities
- Own a research agenda to improve model capability on long-horizon legal reasoning and agentic workflows.
- Design and run post-training experiments using SFT and RL on leading open-weight models.
- Develop reward functions and verifiers for semi-verifiable tasks, including rubrics, learned preference models, and expert preference data.
- Run and debug training at scale, addressing divergence, reward hacking, and instability.
- Build and use rigorous evaluations to measure improvements and determine what ships.
- Design, implement, test, and debug code across the research stack.
Requirements
- Hands-on experience post-training LLMs with SFT and RL, including preference optimization, with direct ownership of training runs and results.
- Deep machine learning fundamentals and experience making distributed training runs work reliably.
- Strong Python engineering skills and experience writing production-quality code with end-to-end system ownership.
- Research experience in reinforcement learning for language models or reward modeling, with the ability to move from experiments to shipped improvements.
- Publications are welcome but not required.
Tech Stack
Categories
AI ResearchML Engineering
About Harvey
Harvey builds domain-specific generative AI for legal and other professional services, delivered as an enterprise platform and APIs to automate contract analysis, due diligence, compliance, and litigation workflows. Founded in 2022 and headquartered in San Francisco, it sells to law firms and corporate legal departments on enterprise agreements; investors include Sequoia Capital and OpenAI. Customers include multiple Am Law 100 firms and Fortune 500 companies’ legal teams.