Senior–StaffAIPlatformEngineer

ABOUT KATAPULT LABS Katapult Labs is a nearshore, AI-first software development partner built around top LATAM engineering talent. We don't do "staff-aug, send a résumé." We embed senior builders directly inside our clients' teams in their meetings, on their comms and back them with the wider Katapult engineering org for the hard problems. We work as one team with our clients: US-holiday aligned, no "us vs. them," and every placement comes with a no-cost replacement guarantee. AI is a first-class tool in how we build, not an afterthought. THE ROLE We're hiring a Senior–Staff AI Platform Engineer to be the first dedicated engineering hire embedded inside one of our clients a US-based, early-stage AI startup building AI for deal and contract workflows (starting in real estate and expanding toward a broader contracts/HR platform). The product is AI-heavy: a chat-first interface layered over document review, deal timelines, and next-step recommendations, powered by a mix of LLMs and classic/deterministic ML. This is not a research or pure model-training seat. We're looking for a builder with high agency who ships production software and operates LLM/ML systems in the real world. You'll build day-to-day with AI coding tooling (Claude Code, Codex CLI) that's how the team works and you'll own the reliability, observability, and quality of AI features in production. KEY RESPONSIBILITIES - Design, build, and maintain scalable ML, AI, and data-platform services that power production features end-to-end. - Operate LLM applications in production: chat with memory, retrieval, prompt management, versioning, and experimentation. - Build and run structured evaluation pipelines (golden datasets, regression checks) so changes ship with confidence. - Own AI/ML observability and monitoring instrument, trace, and debug model behavior in prod with tools like Langfuse or MLflow (knowing one is enough). - Integrate and serve LLM-agnostic model backends (e.g., Anthropic models via Bedrock a