Careers

Build AI that actually ships.

Small, senior team. Real production systems for clients worldwide. If you would rather own the hard last mile than demo a toy, read on.

Agents, LLM apps, evals

AI Engineer

Build and harden agentic systems that run in production for real clients. You will own the parts that make AI trustworthy: evals, guardrails, monitoring, and the human-in-the-loop layer.

What you will do

  • Design and ship LLM and agent features end to end, from prototype to production
  • Build eval suites that turn accuracy into a number tracked per release
  • Add guardrails, cost controls, permissions and monitoring around models
  • Work directly with clients: understand the workflow, scope the bet, ship the system

What we look for

  • Strong Python or TypeScript, and comfort across the full stack
  • Hands-on experience taking an LLM or ML feature to production
  • A bias for measurement over vibes, and for shipping over slideware
  • Clear written communication (we work with global clients, async)
Customer-facing, embedded delivery

Forward Deployed Engineer (FDE)

Embed with our clients and take AI systems the last mile: into their data, their systems and their daily operations. You sit between our engineering, the client teams you work alongside, and real business value, and you own the messy last mile most engineers never see.

What you will do

  • Embed with client teams and take AI systems from prototype into their real production workflows
  • Customize and integrate agents, models and product into each client stack, data and constraints
  • Own the last mile: deployment, rollout, adoption and a clean handover
  • Translate business problems into technical scope, and technical reality back to the people who decide

What we look for

  • Strong full-stack engineering (Python or TypeScript) and speed in unfamiliar codebases
  • Customer-facing maturity: you can hold a room with client engineers and their executives
  • A track record of shipping and integrating software inside someone else's environment
  • Sound judgment under ambiguity, and clear written communication (global clients, async)