Building specialized AI systems that actually understand how work gets done.
Founding Engineer
A small, well-funded team launched this year is rethinking what it means to give every job its own AI. Not chatbots layered on top of existing tools—purpose-built agents that reason about domain-specific workflows, constraints, and edge cases most models miss. You'll design the core orchestration layer that coordinates multiple specialized models, handles ambiguous real-world inputs, and degrades gracefully when things go sideways. This means writing a lot of Python, sweating the details on evaluation pipelines, and occasionally throwing out assumptions when the data tells you something surprising.
What they're looking for
- 2–5 years building production systems in Python, with a strong instinct for when to refactor and when to ship
- Comfort working close to the metal of modern LLMs—prompting strategies, retrieval architectures, and evaluation methodologies are second nature
- Experience designing fault-tolerant distributed systems or agentic workflows that handle messy, real-world data
- Opinionated about developer experience and API design, with examples of interfaces you've built that other engineers enjoyed using
- Excited by ambiguity and the pace of a seven-person team where your decisions shape the product from day one