About Hilbert

Hilbert is a scalable, data science-first growth engine that gives B2C teams predictive clarity into user behavior, revenue drivers, and the actions that drive sustainable growth. Fully agentic by design, Hilbert shrinks months-long decision cycles to minutes.

About Hilbert

Hilbert is a scalable, data science-first growth engine that gives B2C teams predictive clarity into user behavior, revenue drivers, and the actions that drive sustainable growth. Fully agentic by design, Hilbert shrinks months-long decision cycles to minutes. From Fortune 10 enterprises to beloved brands like FreshDirect, Blank Street, and Levain Bakery, operators run their growth on Hilbert. We’re also co‑building alongside leading AI companies.

Position Overview

Hilbert is building a reasoning engine that must navigate non‑deterministic user behavior across data silos. Our demand intelligence platform doesn’t just call APIs. It solves the hard problem of orchestrating multi‑step inference over messy, high‑stakes enterprise data where deterministic answers don’t exist. We’re looking for an AI Engineer who can build production‑grade AI systems end‑to‑end, from prototype to pipeline to product, with the ownership and urgency of a startup culture. This is not a “wire up a prompt chain and move on” role. You’ll own core pieces of the AI stack that power Hilbert’s platform: designing agent architectures, building evaluation systems, and making hard tradeoffs between accuracy, latency, and cost in production. You’ll ship fast while the spec is still evolving, and you’ll explain what you’re building (and why) clearly to the rest of the team. If you think in systems, have opinions about how agentic workflows should actually work,



and want to build AI products that drive real enterprise outcomes, we want to meet you.

What you’ll own first: evaluation and testing for our agents

Hilbert’s agents are in production with enterprise customers today. Before we expand what they do, we need to know, reproducibly, when a change makes them better or worse. You’ll design the eval harness, define what "correct" means for a multi‑step agent trajectory, build the regression gates that run before anything ships, and turn production failures into test cases. From there, your scope widens to retrieval, orchestration, and execution across the AI stack.

What You’ll Do

- Own the evaluation layer for our agents: harnesses, metrics, golden datasets, regression gates, and human‑in‑the‑loop review.
- Architect and implement agent workflows using LangChain, LangGraph, or equivalent, including state memory, routing, tool registries, and recovery paths.
- Own systems from experimentation through production, and operate them once they’re live: tracing, monitoring, latency, cost‑per‑task budgeting, and on‑call for what you build.
- Diagnose real production failures (hallucination, tool misuse, retrieval misses, silent degradation) and turn each one into a durable fix and a test.
- Set the technical standard for agent work at Hilbert: review designs, define patterns others build on, and raise the bar across the team.




- Collaborate closely with the founding team and cross‑functional partners, communicating tradeoffs, progress, and technical decisions clearly.
- Make pragmatic engineering decisions under ambiguity: ship, learn, iterate.

Our current hurdles: the kinds of problems you’ll walk into on day one

- Intelligent retrieval across different approaches. Our agents need the right information at exactly the right moment. The challenge isn’t picking one retrieval method. It’s combining RAG, graph‑based retrieval, and other approaches into one strategy that fetches the most relevant content exactly when the agent needs it, no more and no less.
- Agentic workflows that solve real‑world problems. Workflows have to be robust enough to handle the unexpected. When an agent hits an edge case, missing data, or a situation it wasn’t designed for, it needs to reason through it: use the context it has, elevate to a human when it can’t, and never fail silently.
- Evaluation beyond vibes. We need systematic, reproducible evals that actually predict real‑world performance. If you’ve built custom evaluators for RAG or agent workflows, we want to talk.
- Execution and real‑world integration. An agent that only surfaces insights isn’t enough. We’re building systems where agents take action: integrating with external platforms, executing workflows, and doing real work with the information they have, with human‑in‑the‑loop checkpoints that keep enterprise trust intact.

Who Thrives In The Role

We care about what you’ve shipped and how you think. We expect 4+ years of building production software, with at least 2 of those on LLM or agentic systems that real users depend on.

Must-haves

- 4+ years of pro

📌 AI Engineer (Barcelona)
🏢 Hilbert
📍 Barcelona