Navan’s Cognition team builds and operates AI-powered travel and expense experiences used by customers around the world. We are developing intelligent agents that help users search, make decisions, and complete complex workflows through natural, personalized interactions. Our team owns the AI platform end to end - from agent execution and LLM orchestration to state management, integrations, evaluation, observability, and production reliability. We are looking for a Senior Software Engineer to own and evolve the runtime platform that powers Navan’s production AI agents. This is a hands-on backend and platform engineering role for someone who combines deep TypeScript expertise, strong distributed-systems fundamentals, and exceptional production debugging skills with a practical understanding of LLMs and agentic systems. You will work closely with engineers building AI agents. Your responsibility will be to provide the reliable runtime, infrastructure, abstractions, and observability they need to deliver new capabilities safely and quickly. What you’ll do Own and evolve the production runtime responsible for executing and orchestrating AI agents. Design platform capabilities for agent execution, tool calling, streaming, state management, persistence, and long-running workflows. Build resilient integrations with multiple LLM providers and model-serving platforms. Design provider-routing and fallback strategies based on availability, latency, quality, and cost. Implement retries, timeouts, circuit breakers, rate-limit handling, idempotency, and graceful degradation. Ensure the platform remains available when external dependencies or infrastructure components experience outages. Build reliable mechanisms for loading, caching, versioning, and recovering agent configurations and artifacts. Create end-to-end observability for AI requests, including model, provider, agent, latency, token usage, cost, errors, retries, and fallback behavior. Define dashboards, ale
