Software Engineer, Agentic Workflows

ABOUT REBAR Rebar is building the AI operating system for commercial HVAC, Electrical, and Plumbing. Over the past year our quoting platform has processed tens of thousands of projects across North America and we’re continuing that growth. Our customers include many of the top firms in the industry. Some of these companies are running billion dollar construction projects on workflows that still look like it's 1985. Construction is 10% of GDP and still massively underserved by software. We are changing that. We recently raised a $14M Series A from leading construction tech investors and are entering our next phase of growth. We are building a set of AI native products that will define how this industry operates. ABOUT THE ENGINEERING ORG We're in the age of ai and the role of the engineer is rapidly changing. We're aware. We're being very intentional of ensuring we adapt with it. We are fostering an engineering culture of growth and development. We strongly emphasize care of craft and winning together. ABOUT THE ROLE We're looking for a talented Software Engineer to build the platform and context layer for our Rebar Agent. This role will include utilizing the frameworks, services, and tooling that make agentic workflows reliable, scalable, and genuinely useful in production. This is a hands-on role with broad impact: you're about to build the backbone for an agent, understand a user’s current context, access the right data safely, and take actions with strong guardrails. You'll own this from zero: context modeling, retrieval and indexing, orchestration and tool-calling architecture, evals/observability, and cost + latency. This is a high-leverage product surface for the company, and it's a full-stack role - you'll be writing TypeScript on the UI in the morning and Python on the orchestration layer in the afternoon. You'll join a small, highly capable engineering team building the future of cutting edge construction tech. RESPONSIBILITIES - Design and maintain backend services and orchestration systems that power agentic workflows and tool execution. - Build the context layer - indexing, retrieval, and structured representations so the agent can ground itself in the right information without re-reading everything. - Integrate LLMs into product surfaces safely - permissions, guardrails, and acceptance-gated write actions - Stand up evals + observability - measure quality (accept/reject/edit-then-accept), reliability, and regressions; iterate quickly. - Collaborate cross-functionally with ML and product to bring new capabilities into production and improve developer/customer experience. WHAT WE'RE LOOKING FOR We’re looking for passion and excitement around building AI infrastructure, backend, and clean ux. Our users need to be able to trust and have an intuitive UX so that they can operate smoothly and efficiently in our platform. We want that product sense coupled with excitement and technical prowess. Ideally, you have shipped an agentic experience or product before - you’re aware of failure modes, measure quality quantitatively with evals, and you have strong opinions about agent tools, MCP, and agent responsibilities. You should be fired up about building AI native interfaces that real users trust to read and mutate. You should have experience with document retrieval, knowledge graphs, evals and pertinent productionalization of agentic products. You should also be comfortable owning a feature from the chat UI through the orchestration and tool calling. REQUIRED QUALIFICATIONS - 6+ years of industry software development experience - Full-stack chops comfortability - Deep familiarity with LLM tool use / function calling - Evals and iteration, not just ripping vibes - Comfort with ambiguity - Product taste, especially for trust NICE TO HAVE - Built LLM eval harnesses with acceptance / reject / edit-then-accept tracking, or similar trust metrics - RAG over large document corpora (construction plansets can be 300+ pages and very de