AI Product Engineer

We are building the layer that lets AI agents actually connect to and use any system CRM, ERP, internal tools, systems with no API at all. Every AI agent platform today either has shallow breadth, narrow depth, or proprietary generation with no compounding registry. We're building the connector lifecycle layer underneath all of them: generation, mechanical verification, drift detection, and the hardest unsolved piece promoting browser-automated tasks into native, verified connectors automatically.

The entire OS for ecommerce.

WHAT YOU'LL DO:

Sit with us directly and turn fuzzy pieces of the connector lifecycle (discovery, connection, verification, monitoring, auto-generation) into scoped, shippable product — no PM in between.
Design and build the core technical loop: browser automation for systems with no usable API, reverse-engineering the underlying network calls, generating native connectors from them, and mechanically verifying the native version actually does what the browser version did.
Build full-stack: a Python/Node connector-generation backend, registry data model, lightweight tooling to inspect connector health, and LLM integrations for parsing docs/specs and reasoning about API behavior.
Deploy and own what you ship — CI/CD, monitoring, and you're the one who finds out (and fixes it) when a generated connector breaks in someone's production agent.
Iterate in the open — ship the ugly version first, share what broke, kill the parts that don't work without ego.
Have a real say in sequencing: which APIs we verify first, when the registry goes live, when browser promotion is solid enough to ship.

TECHNICAL DEPTH WE EXPECT:

Browser automation & reverse engineering. Fluent with Playwright or Puppeteer; comfortable reading network traffic (XHR/fetch calls, auth headers, tokens) and turning what you observe into a clean, direct API client.
AI/LLM. You build with LLMs — using them to parse documentation, infer schemas, generate and self-check connector code — not just call a chat endpoint.
Backend. REST API design, schema validation, sandbox/test environments, Python or Node.
Verification instinct. Your default question is "how do I prove this mechanically," not "it looks right to me."
Deployment. Docker, basic CI/CD, you ship your own code to prod without waiting on infra.
Data literacy. Comfortable defining and reading things like pass/fail rates, health scores, drift signals.