The posting
We're looking for a senior or staff-level software engineer to build the AI systems that help the world's best and brightest build in the U.S.
You'll own the agents and infrastructure that power complex legal workflows: drafting documents, reviewing evidence, and moving cases toward filing. This includes workflow orchestration, backend services, and evaluations that measure quality and catch regressions. You'll work directly with our legal team to understand their processes and turn early experiments into reliable production systems.
WHAT YOU'LL DO:
- Build agentic systems that automate complex, multi-step legal workflows, with human review and intervention at the right points.
- Redesign our task and workflow orchestration to support long-running work, dependencies, retries, and recovery.
- Build evaluation systems to measure output quality, catch regressions, and guide improvements to agents, prompts, and models.
- Improve the architecture of our backend services, event pipelines, and scheduled job infrastructure.
- Make agent execution observable, so we can understand what happened, debug failures, and manage quality, latency, and cost.
- Connect these systems to our internal tools and applicant-facing portal, including the APIs and interfaces people need to supervise their work.
YOU MIGHT BE A GOOD FIT IF:
- You have 4+ years shipping production software, with deep experience in backend systems and experience building AI/ML powered applications.
- You've built with LLMs beyond a basic chat interface and thought carefully about tool use, context, evaluation, and handling unreliable outputs.
- You can work with domain experts to turn nuanced processes and quality judgments into software behavior and meaningful evaluations.
- You're comfortable working in Python and TypeScript, or similar languages.
- You've led projects end to end, from ambiguous requirements through deployment and ongoing operation.
- You have a solid understanding of workflow orchestration and asynchronous systems, including queues, events, retries, idempotency, and failure recovery.
- You have high agency: you can independently research unfamiliar problems, test approaches, and drive them to a working solution.
Experience operating agents in production, building eval infrastructure, or working on complex document workflows is especially relevant. Strong engineering judgment matters more to us than experience with a particular framework.
CULTURE
- Engineering is woven into the company. You'll work directly with our legal ops team, product, and customers every week.
- We value independent judgment. When something is unclear, we expect you to dig in, research, prototype, and validate with users before looking for sign-off.
- We protect focused time. Most coordination happens async so people get long stretches of focused work.
- We care about craft. Clean APIs, fast UIs, and well-structured data models matter to us.
- We're bicoastal but tight-knit. The team is split between San Francisco and New York, and we still operate like one small, close team.
PAY AND BENEFITS
- We sponsor visas, obviously.
- Competitive base and equity, scaled with experience.
- Competitive medical, dental, and vision benefits.
- 401k with a 3% company match.
- Equipment/office stipend.
- Annual team off-site in cool places.
- In person in San Francisco or New York.
Lighthouse is an equal opportunity workplace; we welcome people from all backgrounds, communities, and experiences. Apply through our Careers page. If selected, we'll schedule a call to get to know you better and do a deep technical dive into the work you've chosen to highlight.
