AI Agent Developers
You architect, ship, and harden autonomous, tool-using agents that execute real marketing in production — multi-step planning, tool/function calling, durable state, and failure recovery — for thousands of trades businesses at once.
- ✓Architect tool-using agents with explicit planning, function/tool calling, and durable state — not single-shot prompts.
- ✓Engineer reliability: idempotent steps, retries with backoff, timeouts, circuit breakers, dead-letter queues, and human-in-the-loop approval gates.
- ✓Build and version evals — golden sets, LLM-as-judge, regression suites — so every prompt or model change is measured before it ships.
- ✓Instrument every run: distributed tracing, per-step token/cost accounting, and structured logs so you can debug a failed run in minutes.
- ✓Wire agents into the platform, CRMs, ad APIs, and internal tools through typed contracts, queues, and webhooks.
- ✓Own an agent end to end: prototype → eval → canary → production → on-call for its outcomes.
- You've shipped agentic systems to production — orchestration, tool calling, memory, and recovery — not a wrapped chatbot.
- You reason fluently about state machines, failure modes, retries, and idempotency.
- You've used at least one agent framework (LangGraph, OpenAI Agents SDK, CrewAI, or your own) and know its sharp edges.
- You can read a trace and tell us exactly why a run failed — and what you'd change.
- You've carried a pager and pushed a system's reliability up over time.
- LangGraph / OpenAI Agents SDK / custom orchestration
- Function & tool calling, structured outputs
- TypeScript + Python
- Queues, webhooks, idempotency keys
- OpenTelemetry tracing + eval harnesses
- 3+ years shipping production software, with hands-on agentic systems you can demo live.
- A public GitHub and at least one autonomous agent you can walk us through end to end.
- Comfort owning reliability, cost, and latency as first-class concerns.
- Builders, not talkers. You ship products on your own. Your GitHub proves it.
- Dev background first. You earned vibe coding the hard way — real development experience before AI-assisted coding.
- Obsessed with the craft. You can't sleep till it's done, and your feed is constantly full of learning.
- AI infrastructure at scale. A solid understanding is a strong plus — heavy on the back office, or elite at UI/UX.
- A marketing brain. Bonus points if you've built AI automation systems for paid advertising.
- Client-fluent. You work hand-in-hand with non-technical clients — idea to prototype to production, then sell and get results.
- Forward-looking. You see where the world is going, not just where it's been.
- Thrives in fast water. You're built for a fast-changing, ever-evolving environment. So are we.