## The opportunity
LLMs are changing analytical work. Capabilities that once required large teams of highly paid data engineers are becoming accessible to smaller companies for the first time. At Palantir, we delivered real data and BI value, but deployments never scaled without expensive, hands-on engineers. We believe LLMs change that, and we have the customers and revenue to prove the model works.
Now we scale and one key lever is the platform. Every new customer brings new data sources, more rows to sync, and more queries to serve. Your job is to make sure ZeitMind can handle onboarding 10 new enterprise customers per week and the hard part isn't the compute. It's capturing each business's context fast enough: connecting messy data systems, making sure the agent's answers are correct, that visualisations hold up, and that the customer is able to get value out of the product quickly. This is as much a product and correctness problem as an infrastructure one, and it hasn't been solved before. Onboarding a customer should be boring. This is the role that lets everything else scale.
## What you will do
* Build the sync layer: millions of rows from ERP, CRM, and homegrown systems, ingested incrementally and reliably, without an engineer babysitting the pipeline
* Cut onboarding time: connecting a new customer's data sources should take hours, not weeks. You abstract sources so our agents work with any of them the same way
* Make the agent fast where it counts: speed comes from tool design that parallelizes, sub-agents, and branching, not tokens per second. You design tools so work can run concurrently and safely
* Route data safely between customer networks and ours: security and reliability are features our customers pay for
* Build the guardrails for correctness: automatic checks and integrated validation tooling so the agent's output can be trusted, and so it flags what a human should verify
* Keep the platform simple: choose boring technology where boring wins, and be able to say why every system we run earns its place
## You will thrive here if you
* have built or scaled data platforms before and think clearly about data processing architectures
* have a deep understanding of OLAP and OLTP systems and when to reach for each
* bring strong backend experience with TypeScript and are at home in cloud infrastructure
* get foundations right without overengineering them; you build for the scale we will hit next year, not for a hypothetical one
* take full ownership from idea to production to impact, and are comfortable working without predefined specs
* want to build foundations early rather than optimize mature systems
* are genuinely interested in the data and agentic space and how LLMs enable new workflows for non-technical users
## Requirements
* You've been a lead architect or equivalent: built many systems yourself, and seen how large systems fail and evolve. You're here to learn from customers and take bets on a product that doesn't exist yet, not to learn how to write software.
* Strong TypeScript and cloud infrastructure experience
* English C1 or above; German a plus
## What we offer
* We are based in Munich and build the team around in-person work. We pay for your relocation.
* Regular San Francisco offsites for product sprints and staying close to the frontier.
* Zeit AI package: daily lunch allowance, free in-office dinner, wellpass membership, best tech & tools
* You know someone? €5k referral bonus for every successful hire.
## Tech stack
TypeScript end to end. Backend on Bun with Postgres and Supabase. React frontend (Mantine, TRPC, TanStack Query). Job/Process management, Docker, VPN/routing and Agentic frameworks: harness, tool calls, sandboxed tools, branching of data and specs.
## Apply
Still apply if you do not match every requirement. If you are exceptional in core areas and learn fast, we want to talk.
Visa sponsorship is offered for this role.
