Software Engineer, Agents
About The Position
As a Software Engineer, Agents at Harvey, you will build systems that make AI agents indispensable to legal professionals. The work includes designing environments and actions for agentic professional work, selecting models, managing context windows, creating tools, and developing evaluations that enable faster iteration and new capabilities. The role is aimed at engineers and researchers immersed in agentic systems who want to ship impactful products and use practical evaluations to improve task completion quality and customer satisfaction. You will work with customers and product managers to understand legal workflows, define practical evaluations of excellent outcomes, and deliver agents that complete the work. You will optimize agent performance through prompt engineering, model selection, tool design, skill writing, context-window management, and evaluation-harness development. In collaboration with model infrastructure teams, you will design and implement infrastructure for low-latency agent execution, including caching, parallel tool calls, and subagent patterns. You will also improve observability and instrumentation, profile agent behavior, identify bottlenecks, and apply new developments in agentic systems to Harvey’s products. Example projects include automating information requests and diligence checks across large collections of files for M&A deals; improving the latency and quality of agents applying legal playbooks to contracts; optimizing multi-source retrieval agents for litigation; and enabling board-ready PowerPoint generation by tuning coding-agent harnesses and libraries.
Requirements
- Passion for building effective domain-specific agents.
- An iterative mindset: develop proofs of concept, make decisions quickly, and ship early versions.
- Comfort with using evaluations to drive quality.
- Adaptability with code and frameworks, and willingness to drive adoption of new best practices.
- 2–5 years of software engineering experience after a bachelor’s or master’s degree.
- Proficiency in Python and experience working with LLM APIs and agent frameworks.
- Experience shipping user-facing products on the backend or as a full-stack engineer.
Responsibilities
- Partner with customers and product managers to understand legal workflows, define practical evaluations of excellent outcomes, and ship agents that complete the work.
- Optimize agent performance through prompt engineering, model selection, tool design, skill writing, context-window management, and evaluation-harness development.
- Work with the model infrastructure team to design and implement infrastructure for low-latency agent execution, including caching strategies, parallel tool calls, and subagent patterns.
- Improve observability and instrumentation to profile agent behavior, identify bottlenecks, and guide optimization decisions.
- Stay current on developments in agentic systems and apply relevant learnings to Harvey’s products.
Stand Out From the Crowd
Upload your resume and get instant feedback on how well it matches this job.
What This Job Offers
Job Type
Full-time
Career Level
Mid Level
Education Level
No Education Listed
