We’re hiring an Applied AI Engineer to push the boundaries of our Cofounder agent. You’ll own core backend systems and applied LLM work: advancing agent reliability and autonomy, building evaluation pipelines, and shipping techniques that measurably improve agent performance. This is a hands-on role with high ownership across research-to-production: prototyping, instrumenting, evaluating, and deploying improvements that show up directly in user outcomes.

What You’ll Do

• Design and implement agent improvements end-to-end: prompting strategies, tool selection, action planning, memory usage, safety/guardrails, and recovery paths
• Build robust evaluation pipelines for the agent: offline evals (golden tasks, regression suites, behavior tests), online metrics (latency, success rate, fallout modes, cost efficiency), and experimentation frameworks (A/B, canaries, guardrail thresholds)
• Productionize applied LLM techniques: function/tool-calling orchestration, self-reflection, retrieval/RAG, multi-agent handoffs, caching/embedding strategies, and hallucination reduction
• Improve core backend systems: reliable job orchestration, retries/backoff, idempotency, and auditability; scalable memory and context routing; data pipelines across Gmail, Slack, Notion, Linear, Google Workspace, etc.; observability and tracing for agent actions/outcomes
• Partner with product and infra to define success metrics and ship fast, safe iterations
• Write clean, well-tested code; document design decisions and runbooks

What You’ll Bring

• 4+ years backend engineering experience, preferably Python (we care about impact over years)
• Hands-on LLM experience: prompt engineering, function-calling, retrieval, embeddings, evaluation design; you’ve shipped LLM features to production
• Track record building evaluation harnesses and using them to drive improvements (regression suites, task success metrics, cost/runtime tradeoffs)
• Solid distributed systems fundamentals: concurrency, reliability, performance, data modeling, lifecycle management
• Pragmatic experimentation: hypothesis → prototype → measured improvement → rollout
• Excellent debugging and instrumentation skills; you enjoy finding and fixing edge cases in the wild

Nice To Have