Founding Engineer - AI
7 hours ago
Barcelona
You build the LLM reasoning that runs underneath everything we ship: turning messy logistics data, contracts, invoices, claims, tickets, into clean, canonical structure that machines can act on. This is a horizontal role: your work powers every product surface, not one. If you think software-first, AI-second, this is your seat. (Prefer owning batch pipelines and dashboards? Look at our Software Engineer opening instead.) $2.5M pre-seed, backed by Seedcamp and Yellow alongside top-tier investors in London, Paris, and Madrid. On-site in Barcelona, moving fast. You'll work in a small, high-agency team where you own problems end to end and see your code hit production the same week. We're building the Logistics Brain. Every year the logistics industry loses a fortune to invoices, contracts, and claims still handled by hand. Carriers like UPS, FedEx and DHL overcharge and mishandle shipments at scale, and their customers rarely have the tooling to catch it. We turn that mess into a reasoning engine that audits invoices line by line, recovers money, and closes carrier claims automatically — and our customers watch the money come back. • Design LLM reasoning pipelines that extract and normalize unstructured logistics data — documents, contracts, invoices, claims, tickets — into the canonical schemas the whole product runs on., • Build and tune RAG + retrieval so extraction stays accurate on real, ugly documents: many languages, bad scans, inconsistent formats., • Engineer Human-in-the-Loop feedback loops that turn operator corrections into measurable evals and model improvements., • Ship agents into production. Our claim and ticketing agents run on Temporal with PydanticAI workers making real decisions on live data., • Own the evals. You define what "good" means before you ship, then prove it., • Keep cost and latency in check as volume grows: batching, caching, and picking the right model for each task., • 2-3 years building software, with real production experience shipping LLM/ML features., • A software-first approach to AI: you care about reliability, latency, cost, and evaluation, not just prompts., • Experience with data normalization and reasoning over unstructured documents., • Power user of AI coding tools, using agents and plugins to keep an extreme execution pace., • OCR cleanup and document translation at scale., • Deep vector DB and retrieval optimization., • A hacker instinct for making LLMs reason through domain-specific logic., • Meaningful equity. We're all owners of the Logistics Brain., • On-site in Barcelona., • AI budget for every tool you want, • Top-spec hardware (MacBook), • Off-sites in Spain and beyond., • Founder/engineering chat - intro, agency, ownership, the Logistics Brain., • Live technical challenge on a real case (1h). Project-style, get something working. You'll use AI tools., • Team fit & final sync - trade-offs and long-term vision. #J-18808-Ljbffr