Product Engineer - Intern

## About the role You build the applications customers actually use. Compound turns messy business operations into AI-native systems: software that understands how a company works, connects across its tools, and gives people agents that can actually do work. That means our product engineers are not just building dashboards or wrappers around models. They are building the interfaces, workflows, data systems, orchestration layers, and evaluation loops that make AI useful inside real companies. This is a full-stack role for someone who likes owning the whole thing: product judgment, frontend, backend, data, LLM orchestration, observability, and production reliability. You should be equally comfortable debugging an agent loop at 2am and deciding whether a workflow makes sense for an ops team the next morning. ## What you own You own AI-native applications end-to-end. - React and TypeScript on the frontend - Python, FastAPI, Postgres, and background jobs underneath - LLM orchestration in the middle - Evals, observability, permissions, integrations, and the unsexy plumbing that keeps the whole thing running You will take features from prototype to production to iteration. Some days that means building a new customer-facing workflow. Other days that means designing an eval harness, fixing latency, improving data ingestion, or turning a fragile demo into a reliable system. ## What you bring You have shipped production software across the stack. You can move fast without creating a mess. You have strong opinions about product, systems, agents, RAG, evals, and reliability because you have actually built things. You do not need a PM to translate a stakeholder's problem into a system. You can talk to a user, understand the workflow, make the right tradeoffs, and build the thing. ## Strong signals - Production experience with React / TypeScript and Python - Comfort with Postgres, APIs, background jobs, and async workflows - Real experience with LLM APIs, agents, RAG, evals, or AI-native applications - Ability to debug across the entire stack - Taste in product and UX, not just implementation - High agency in ambiguous environments ## Not a fit if - You need clean tickets before you can start. - You only want to work on frontend or only want to work on backend. - You think AI products are mostly prompt engineering. - You are uncomfortable with customers, messy systems, or fast iteration. ## Why this role is different Most AI products die between the demo and production. Your job is to close that gap. You will build systems that go into real businesses, touch real workflows, and change how teams operate. The work is technical, but the bar is not "does the model respond?" The bar is "does this system become part of how the company runs?" ## How to Apply Send us a short note and a few examples of what you have built. We care much more about evidence than polish. The best applications usually include: - A brief explanation of why Compound is interesting to you - Links to products, projects, repos, demos, or writing you are proud of - A short description of the hardest thing you have built - Anything that shows how you think, ship, debug, or learn You do not need a perfect resume. You do need to show that you can build, figure things out, and operate with unusual ownership.