AI Engineer, Haddock

• 👉 1+ years of hands-on experience building AI systems or automations that real people use every day
• 👉 You're truly AI-native: you follow new models, frameworks, and techniques, and you continuously refine how you use them in practice
• 👉 You're also AI-native in how you build: you use modern AI-powered dev tools (Claude Code, Cursor, Codex or similar) and assemble your own agents and workflows to move faster
• 👉 You think in systems, not tasks. When you watch someone do something by hand, your instinct is to ask what the process really is, and then whether it should exist at all
• 👉 You're comfortable with product engineering. TypeScript or Python, and you're comfortable with APIs, webhooks, jobs, queues, databases, and shipping a frontend when the workflow needs one. The prompt is often the smallest part of the problem, and you own the service around it rather than handing it to someone else to maintain
• 👉 You have business empathy. You can sit with a CX agent or a salesperson, understand their workflow, and turn it into a system. You enjoy talking to people, not only to a terminal
• 👉 You're analytical and quantitative: you measure before optimizing, but you never let analysis slow down shipping
• 👉 Fluent in Spanish and comfortable in English. Most of your stakeholders work in Spanish, so this one matters

Desirable

• This is a x10 role. Not the kind that makes one team 10% faster. The kind where an agent you ship on a Tuesday takes a full week of manual work off an entire department, every week, from then on
• ➡️ Why are we hiring?
• Haddock grows fast: more restaurants, more invoices, more conversations, more deals, more partners, and with all of it, hundreds of opportunities to leverage AI every single week across CX, Sales, Marketing, Partners, Finance and People
• The team is growing fast too, but the operational load grows faster. Rather than let every new process turn into another pair of hands, we're building an AI team focused on internal operations, designed to hyperscale the company with agents
• The team is small on purpose: you'd be joining Héctor, with Enric (AI Lead) as your manager. Small enough that what you ship is visible company-wide within days
• ➡️ Where we are today
• Some examples of what is live:
• Rosetta, an internal agent that answers any question about our product or our customers' data. It takes its context from the codebase and from production, not from the docs, finds the evidence, and explains it in plain language to non-technical people. It lives in Slack and behind our internal MCP, so anyone can reach it from wherever they already work. It never goes stale, because production is its source of truth
• An agent that watches every change in the codebase and turns it into the product changelog our customers read. A daily job goes through merged PRs and our product-news channel, works out what is genuinely customer-facing, groups related PRs into a single item, and drafts the release note. A human reviews and publishes. Prompts, versions and traces live in langfuse
• Our own MCPs over third-party tools. When a vendor's MCP doesn't expose what we need, we don't wait for their roadmap. We build ours: exactly the tools our agents need, with the permissions we want, shipped at our pace. It's the difference between an agent that can look something up and an agent that can actually act on it
• The harness behind the AI cowork tool every employee uses. Org-wide context, an internal plugin carrying our curated knowledge and skills, an MCP gateway exposing scoped internal tools over OAuth, plus the rollout, the training and the adoption metrics for the entire company
• That's the starting point, not the finish line. Support, onboarding, finance operations, reporting and most of what People and Marketing do by hand are still on the list
• ➡️ What you'll do:
• Hunt for leverage. Sit with CX, Sales, Marketing, Partners, Finance and People, understand their real day, and find the processes that eat the most hours. Then decide what's actually worth automating. Some problems need a multi-agent system, some need 20 lines of code, and telling them apart is most of the job
• Build the agents and own the software they live in. This role is not limited to the AI layer. An agent is only as useful as the harness around it: the triggers, the tools it can call, the data it can read, the service that runs it, the interface where a human reviews its work. You own all of it, from the first commit to production and to whatever breaks at 9am on a Monday
• Own the systems, not just the integrations. HubSpot, Zendesk, Slack, Notion, our own product databases, GCP. Sometimes that means migrating or reconfiguring a tool before anything intelligent can run on top of it, and when the integration you need doesn't exist yet, you build it. Whoever owns the system decides what the AI on top of it is allowed to do, so we'd rather that be us
• Own AI observability for everything you ship. Traces, evals, metrics. An internal agent nobody measures is an internal agent nobody trusts
• Maintain and improve our own AI frameworks. We build our own tooling for AI observability, evals and human annotation instead of settling for whatever comes out of the box. You'll use it every day, and you'll be one of the people making it better
• Prove your impact with numbers. We keep a ledger of automated workflows with a measured baseline before and after. We don't declare hours saved, we measure them, and that's on you for whatever you build
• Do your own project management: you find the problem, you sell the solution internally, you roll it out, you train the team that will use it, and you keep it alive
• Build the internal platform: MCPs, skills and shared tooling, so every haddocker can build their own small automations on top of what you ship
• ➡️ Our tech stack:
• TypeScript, Python, langfuse, PostHog, Google Cloud Platform, mastra, Postgres, MongoDB, and several LLM providers
• On top of that, the systems the company runs on (HubSpot, Zendesk, Slack and Notion) and the MCPs, skills and agents we build over them

• ➡️ Our Hiring Process:
• We move fast and communicate openly. Our goal is to wrap up the process within two weeks
• Intro call with Enric (AI Lead), 25 mins
• Call with Pol (co-founder, CPO), 25 mins
• Technical interviews with Enric and Guillermo (Head of Tech), 2 hours
• Final call with Arnau (co-founder, CEO), 25 mins
• Sounds like an opportunity? Apply now. We'd love to meet you