AI Agent Engineer

We're looking for a production software engineer who builds agentic systems — and ships them into live, revenue-bearing environments. What you'll build: The agentic framework that orchestrates decisions across the lead funnel — tool use, multi-step workflows, and the logic that ties them together into a reliable system. The integration layer that wires our decision and pricing models into a live production pipeline, so the optimization logic actually runs against real traffic. The reliability layer that production demands: latency management, error handling, retries and fallbacks, observability, and graceful degradation when something upstream fails. Evaluation and testing for non-deterministic systems — how you verify an agentic workflow behaves correctly when the same input doesn't always produce the same output, and how you catch regressions before they hit revenue. The deployment and monitoring setup that lets us ship changes safely and see what the system is doing in production at all times. What we're looking for: Must-haves: 5+ years total software engineering, with 2+ years specifically building LLM/agentic systems. Strong production software engineering — you've built, shipped, and operated real systems that something important depended on. Built LLM-powered agents or agentic workflows in production, not just prototypes or demos Hands-on with modern agent frameworks (LangGraph, AutoGen, CrewAI, Letta, or similar) Experience with orchestration, tool use, and evaluating/testing non-deterministic systems Strong Python (TypeScript a plus) Solid instincts for reliability: observability, error handling, latency, and failure modes Strong signals: Founding or staff engineer experience at an AI/agent company (think Cognition, Adept, Imbue, Reflection, Lindy, Decagon, Sierra, Cresta, or similar) Open-source contributions to agent frameworks (Cline, LangGraph, AutoGen, CrewAI, Letta) Experience integrating ML models into live production systems Not what we're after: No-code / automation-only builders (Zapier, Make) and GPT-wrapper work Prompt engineering as the primary skill Demo-only agent experience with nothing shipped to production