TL;DR We're building the SAP/NetSuite of the AI era for mid-sized companies in the real-world economy, starting with their supply chain operations (buy, sell, stock, produce) Senior/staff-level founding role : you'll architect the agentic layer and the data model beneath it, not execute a backlog €3M Pre-seed (Top-tier VCs) World-class peers, massive impact Competitive salary + Equity Remote-first What this role is (and isn't) This is a systems-engineering role where the LLM is one component, not the whole job. Half of it is backend and data design: schemas, permissions, write paths, workflow infrastructure. The other half is agents: tools, context engineering, evals, cost and failure modes. What you'll build An extensible ontology. Piston runs on a customer-extensible data model: every client reshapes it to their business, and agents must reason over schemas they've never seen. You'll design the abstractions that make that possible and safe. Agents that act, safely and with supervision. Our agents are not just assistants that answer questions, they act on live business data, under the same rules and permissions as human users. You'll design the framework in which agents act and are supervised. Agentic processes. Agents become part of each company's processes, automating the steps where they're reliable and handing off where they're not. The boundary moves as they prove themselves. Client-specific evaluation. Each company has custom processes and we'll need to be able to evaluate our agents in these ad hoc cases. You'll design evaluation that's grounded in each client's own data and outcomes, giving them access to continuous improvements backed by numbers. Our stack: Python, Temporal, pydantic-ai. What you bring You love deep-diving into ambiguous, complex problems and surfacing with a production-ready solution for real users and challenges. Must have 5+ years of experience building production software; expert in at least one language, autonomous in Python You've owned the architecture of a system: data models and abstractions other engineers built on top of You've shipped LLM-powered features to production and lived with the consequences: evals, cost, latency, failure modes Strong relational-database fluency: schema design, constraints, permissions Nice to have You've built and operated agents in production: tool use, multi-step orchestration, autonomous actions Experience with agent frameworks and SDKs: pydantic-ai, LangGraph, Claude or OpenAI SDKs Python web experience. Django or FastAPI a plus You've worked on metadata-driven platforms: ERP, CRM, low-code engines What you'll get Competitive salary + meaningful equity World-class peers Massive impact on a product shipping to real businesses Regular company offsite (2–4 times per year) in great locations Ideal equipment and setup Remote-first Process We value efficiency and clarity. The entire process takes no more than 2 weeks: 30-minute intro call with Eric (CEO) 90-minute technical interview with Eugène (CTO) 30-min cultural-fit interview with Florian (CRO) 2 days paid assessment Offer: If it's a match, we move fast. Interested? Tell us why you want to join. A short, personal note goes a long way, AI-generated cover letters don't.