AI Software Engineer

Full-Stack Engineer PositionOneTrack is a profitable, growth-stage technology company redefining manufacturing and logistics with physical AI. OneTrack operates as the intelligence layer for physical operations. We serve the world's largest brands and logistics companies by deploying camera sensors and other data-capture devices alongside our mission control platform and AI agents. Today, OneTrack runs in hundreds of warehouses and factories across North America with companies like Ashley Furniture, Mars Snacking, The Wonderful Company, Ryder, and Church & Dwight.We're hiring a full-stack engineer to join our team and work on AiOn, our agentic AI platform. The scope covers the whole product: the UI operators use, the agent and prompt design behind it, the Python services, and the AWS infrastructure they run on. From floor-level to supervisors to C-suite executives at Fortune 500s, our agentic AI platform connects the systems of record with places of action in some of the most demanding operational environments.AiOn is already live in hundreds of warehouses and factories, so the work splits roughly evenly between new features and making what exists faster, more reliable, and more extensible. We integrate frontier models from multiple providers rather than train our own, so the engineering is in the product around them. We work from first principles and ship the simplest thing that solves the problem, then refine it.ResponsibilitiesShip and evolve full-stack features from UI through API to agent logicBuild and extend our agent framework: tool and function calling, multi-step reasoning, and context managementBuild structured tool and data access over operational data with schema grounding, validation, and safe execution, and integrate the external systems agents act onBuild evaluation and observability for quality, latency, safety, cost, and product outcomes such as task completionImprove reliability, robustness, and latency across a distributed, event-driven backend, and re-architect early implementations as the platform maturesMaintain guardrails: least-privilege tool design, safe execution boundaries, fallbacks, and kill switchesWork anywhere in the stack the problem requires: React and TypeScript UI, an Electron desktop app, streaming interfaces that stay usable while output is still partial, and Python services on AWS built around async and event-driven patterns