Building AI agents that turn unstructured documents into structured, queryable workflows for the enterprise.
AI Research Engineer
We’re working on giving large language models the ability to read, reason over, and act on complex business documents—think messy PDFs, scanned contracts, regulatory filings. This isn't about wrapping an API around a chat interface; it's about teaching models to perform multi-step document OCR, extraction, and decision-making with near-perfect reliability. You’ll design and run experiments that push the boundaries of retrieval-augmented generation, agentic reasoning, and evaluation methodology. Your research will ship directly into products used by engineering teams at some of the largest companies in the world.
What they're looking for
- 3–7 years of hands-on experience building and shipping deep learning systems, with fluency in Python and PyTorch.
- Demonstrated ability to design rigorous experiments, analyze failure modes, and iterate on model behavior in production contexts.
- Comfort owning the full lifecycle—prototyping in notebooks, containerizing with Docker, and deploying on Kubernetes.
- Strong communication skills suited for a research-driven engineering culture; you write clear design docs and present findings to technical peers and leadership.
- Prior work on document understanding, OCR pipelines, structured extraction, or agent-based reasoning is a significant advantage.