- Invent the agent roadmap by collaborating with sales, data, and product teams to identify and prioritize high-impact automation use cases.
- Design and build agent architecture, including tools, loops, sub-agents, memory, state, and evaluation frameworks.
- Develop and maintain a platform for scalable agent creation, including shared tool layers and MCP integrations.
- Harden WatchChat by optimizing for multi-turn state, latency, cost, reliability, and regression testing.
- Implement rigorous quality measurements using golden datasets, programmatic verifiers, and LLM-as-judge frameworks.