About The Role

ABOUT THE ROLE

This is a high-impact, customer-facing engineering role at an early-stage AI/ML startup where you own the full lifecycle of new customer deployments — from initial integration through first detection to first merged PR. You'll sit at the intersection of product and engineering, reporting directly to the CEO and CTO, and your work will directly shape how the product evolves.

WHAT YOU'LL DO

Own new customer deployments end to end: drive integration, surface first traces, catch first failures, and land PRs in their codebases.
Embed in customer Slack channels and become the go-to expert on their agent architecture and failure modes.
Write detection logic and evals tailored to each customer's specific problems, then generalize recurring patterns across customers.
Open PRs against unfamiliar codebases and defend them in code review with engineers who didn't ask for your input.
Represent the technical team solo in critical conversations including security reviews, deployment architecture decisions, and trust-building discussions.
Feed learnings back into the product to reduce manual work over time.

WHAT WE'RE LOOKING FOR

3+ years shipping production software with demonstrated customer-facing experience (forward deployed, solutions engineering, consulting, founding, or running your own business).
Strong TypeScript proficiency and comfort reading, patching, and writing production code in unfamiliar codebases.
Hands-on experience building and deploying AI agents in production, with real opinions about why they fail.
Experience writing detection logic, evals, and observability/monitoring code for production systems.
Proven ability to rapidly understand new systems and identify critical issues under time pressure.
Comfortable navigating ambiguity — able to walk into an unfamiliar company and figure out what matters within a week.
Experience with security reviews and deployment architecture discussions is a plus.
Bonus: you've personally run agents in production and have the incident war stories to prove it.