We build production agentic systems that read from and write to internal resources. You will design, ship, and measure them.
What you'll do
- Ship end-to-end: Take agentic workflows from prototype to production, build user interfaces around them.
- Design agent architecture: Define tools, memory, retrieval strategies, and human-in-the-loop handoffs.
- Integrate systems: Expose internal tools and data via MCP servers and APIs with least-privilege access.
- Evaluate performance: Build trace-based evals and golden datasets to catch regressions across full agent runs.
- Operate in production: Implement tracing, token budgets, cost controls, and safety guardrails.
- Scope problems: Work with domain owners to identify where agents add value, and push back when a simpler solution is better.




