AI Agents & Automations Internship (f/m/x)

Job description

Our goal is to have a solar system on every roof, a storage unit in every house, and an electric car in every garage. Enpal makes this possible with an integrated total solution for decentralized energy—from solar systems and battery storage to wall boxes, smart meters, and heat pumps. At the heart of it all is our AI-powered platform Enpal.One +, which intelligently connects thousands of systems and efficiently optimizes electricity procurement and feed-in on the energy market. Enpal Energy is Enpal’s electricity division, making clean energy accessible to every household—from Germany’s most affordable electricity rates to smart energy solutions that optimally integrate generation, storage, and consumption. With ambitious growth targets and a rapidly expanding customer base, we are building a team that is actively shaping the German energy market. In doing so, we combine startup agility with the scale of a market leader—across all areas, from operations and technology to finance, customer support, and product development. What you'll do Improve the agents we already run: We have agents in production today. You’ll make them smarter, more reliable, and cheaper. Automate the manual work: There’s a long list of repetitive tasks people on the team still do by hand. You’ll turn those into agent workflows, one by one, and watch hours come back to the team. Own the flow end-to-end: Understand the manual process, design the agent flow, build it, put it in front of the person whose job it touches, iterate until it actually works for them. Debug the messy parts: You’ll dig into traces, figure out why a tool call went sideways, fix the code, and ship the fix. Work close to the business: You’ll sit with ops and tech people to find the next thing worth automating. We ship a solution in days, not months. What you'll bring Studying at a strong university, ideally CS, ML, or engineering. You’ve actually built with AI: Not just prompts in ChatGPT or using AI to generate code. You’ve used the Claude code or Codex , wired up tool calls, handled streaming, and shipped agent teams. Comfortable in code: TypeScript and/or Python. You can read an existing codebase, make changes, and ship a PR without hand-holding. Agent-adjacent experience is a big plus: LangGraph, Claude Agent SDK, MCP, vector DBs, RAG, evals, or any serious side project involving agents. If you’ve built your own…