Build the infrastructure for fair subscriptions Berlin (hybrid/on-site preferred) – EU remote possible Permanent, full-time ----- The Work We're building the platform that secures fair deals on any subscription — energy today, any other subscription type tomorrow. We have PMF and hundreds of thousands of users, but to reach millions, we need to automate operations that are currently manual. You'd be building the platform that makes that possible. This is backend-heavy, systems-level work — the kind where you're holding multiple interacting systems in your head and hunting for the edge cases that will break them. The Context: SwitchUp Tech SwitchUp Tech (https://switchup.tech) builds the technology for two sister organizations: SwitchUp.de (https://www.switchup.de/) , and Fairbraucherschutz (https://www.fairbraucherschutz.de/). Today, our operations teams act as human APIs between users and legacy provider portals. We're automating that so the team can focus on complex cases where humans genuinely add value. The opportunity: You wouldn't be joining a fully-fledged team — you'd be shaping one. Three people today, hiring 6-8 this year, staying small and nimble. No runway pressure — we're funded by SwitchUp.de (45 people, bootstrapped, profitable). The Engineering Challenges The Abstraction Layer Challenge: How do we architect a system agnostic to the Provider (Vattenfall vs. Deutsche Telekom) and the Market (Energy vs. Telco), so we can scale without rewriting core logic? The Robustness/Configurability Paradox: Workflows that are highly configurable (to experiment fast and move to new subscription types easily) but robust enough to handle millions of subscriptions. Human-in-the-Loop AI: Orchestrating workflows where agents start tasks and humans intervene only when confidence drops — with proper failure handling, retries, and rollback. The Stack TypeScript-first, AI-friendly tooling designed to scale to hundreds of millions of workflow instances a year. How We Work No PMs/POs. No specs.*You own a business problem end-to-end — from sitting with operations to build understanding, to architecting the system, to shipping and measuring impact. AI-forward. As new capabilities emerge, we rethink how we work. Lots of room for experimentation and knowledge-sharing — everyone is both student and teacher. If you're energized by colleagues who explore new capabilities together the moment they drop, you'll fit in. If that sounds exhausting, we're not the right place. Who You Are NOT Someone who picked an AI coding tool 6 months ago and hasn't questioned it since A specialist who wants to stay in one corner of the codebase An engineer who prefers others to define the problem so you can focus on implementation Apply No cover letter needed. We want to see how you think. Your CV / LinkedIn / GitHub A system you've built that's similar to ours: complex workflows, state machines, highly configurable products. What was the problem, and what made it hard? What improvements you've recently made to your AI-enabled workflow — what worked well, what's still holding you back?