Mid AI Engineer
hace 2 días
Barcelona
About Valeria Valeria is building the future of HR and payroll in Spain. We’re an AI-native platform that automates contracts, payroll, and compliance for companies with high employee turnover (hospitality, delivery, events, agriculture). We’re rethinking how an entire industry works—moving from manual, error‑prone processes to intelligent automation. ¿Listo para inscribirse? Antes de hacerlo, asegúrese de leer todos los detalles pertenecientes a este trabajo en la descripción a continuación. We’re a fast‑growing startup backed by top investors, disrupting a €5B industry that is still stuck in spreadsheets and legacy software. Your Role You’ll work closely with the AI Lead, designing product AI features and transforming internal processes with AI, helping build a cross‑functional AI team with impact across every department. You’ll be a core member of our AI team, building the agents that power Valeria in production—talking to real customers and handling real payroll and legal processes. You’ll own AI features end‑to‑end: designing agent architectures, engineering context, building evals that hold up, and shipping reliable systems into a domain where correctness genuinely matters. Responsibilities • Prototype the complex: build proofs of concept that solve hard problems in innovative ways, then take them to production, • Translate business into AI: understand the business problem deeply and land it into a solid technical solution, • Design agents end‑to‑end: multi‑agent architectures, tools, tool‑calling, function calling, memory, orchestration, and state management, • Master context engineering: decide what goes into the context window and how (system prompts, few‑shot, retrieval, memory, compaction, token management), understanding why behaviour changes and anticipating failure modes (hallucinations, edge cases, prompt injection), • Build the LLM harness: the layer around the model—tool interfaces, output parsing and validation, retries, fallbacks, guardrails, scaffolding, and flow control—that turns a model into a reliable production agent, • Ensure reliability: solid evals and observability (datasets, metrics, regressions, production tracing) before every release—never on a single happy‑path, • Build high‑quality RAG systems: embeddings, vector stores, chunking, retrieval, re‑ranking, and grounding in a compliance‑heavy context, • Pick the right model: integrate and compare GPT, Gemini, and Claude, reasoning about cost, latency, reliability, context window, and fallback, • Integrate systems: build MCP servers and integrations with external systems, • Ship production code: solid Python, APIs, tests, CI/CD, and the team’s best practices, • Stay on the frontier: keep up with the latest models and technologies and test them to spot opportunities, • Own features end‑to‑end: from technical design to deployment, monitoring, and iteration based on customer feedback, • Mentor interns and evangelise AI across other departments as we scale the teamRequired, • 3 years of professional software engineering experience building production systems, • Strong Python skills and solid backend fundamentals (APIs, SQL, Git, testing), • Hands‑on experience with LLMs / agents in production: LangChain / LangGraph, RAG, prompting, tool‑calling, or equivalents, • Context engineering and evaluation mindset: you reason about why models behave the way they do, and you validate with evals instead of a single test, • Ability to design and break down medium‑complexity solutions autonomously, communicating progress, blockers, and trade‑offs clearly, • Startup mindset: comfortable with ambiguity, high autonomy, and fast iteration cycles, • Strong communication skills and ability to collaborate across product, design, and business teams, • Fluent in English and/or SpanishHighly Valued, • Cloud experience: Azure (Azure OpenAI / AI Foundry) and GCP (Vertex AI / Gemini), • Hands‑on practice / familiarity with AI coding tools such as Claude Code, Cursor, Codex, and similar, • React and basic frontend notions for full‑stack contributions, • Experience deploying agents/models in production at scale, • MCP, advanced function calling, and evaluation frameworks (LangSmith, RAGAS, or similar), • Fine‑tuning / model optimization techniques, • Background in FinTech, HR‑tech, or regulated industries (compliance‑heavy products, government integrations)What makes you a great fit You’re the kind of engineer who treats LLMs as systems to be understood, not black boxes to be prompted once. You care about reliability, you anticipate how agents fail, and you build the harness and evals that keep them honest in production. You’re pragmatic but principled, comfortable moving fast in a startup, and excited to work in a domain where correctness matters—getting payroll wrong affects real people’s lives. Bonus points if you love being on the frontier of applied AI. Our Stack • Language: Python, async APIs, SQL, Git, • AI frameworks: LangChain / LangGraph, • LLMs & agents: prompting, context engineering, agent harness/scaffolding, tool‑calling, multi‑step agents, RAG, embeddings, vector databases, • Models & Cloud: Azure OpenAI, Gemini (GCP), Claude, • Quality & Observability: evals, testing, LLM tracing (Datadog), • Channel: WhatsApp APIOur technical philosophy As an AI‑native product, we build intelligence into every layer—automating altas, bajas, payroll, and compliance through agents that run in production, not demos. We care about reliable, testable systems over framework magic, and we treat evals and observability as first‑class. xcskxlj If you’re excited about applying AI to solve real business problems (not building AI for AI’s sake), you’ll love working here. What We Offer • Competitive compensation: €40,000‑45,000 gross salary Equity, • Free lunch when you’re at the office thanks to Kombo & Nora, • Flexible remuneration with Coverflex, • Flexibility: Hybrid setup (HQ in Barcelona), 60 days/year remote work from anywhere, • Unlimited vacation days—take the time you need, no counting daysOur hiring process, • Intro call with People (30 min), • Interview with the Hiring Manager (45 min), • Tech Assessment – Onsite at the office (1 hour), • Founders interview (45 min), • OfferWhy Join Valeria Now?, • Timing: We’re past the “idea stage” with real customers and revenue, but early enough that you’ll define how we scale our AI, • Market opportunity: €5B market in Spain, every company with employees needs payroll, and current solutions are outdated and painful, • Real AI ownership: you won’t assist on AI projects—you’ll build them, decide on them, and see their impact on thousands of people, • Career growth: be a critical AI hire, build the playbook, and grow as we scale the team #J-18808-Ljbffr