Frontier Agentic AI Engineer — Santander AI Lab

Frontier Agentic AI Engineer – Madrid, Spain

Santander () is evolving from a global, high‑impact brand into a technology‑driven organization. The Chief Data & Artificial Intelligence Officer division is building a world‑class AI & Data team to make a difference for over 170 million people worldwide.

• Design, build and deploy production‑grade agentic AI systems – multi‑agent orchestration with real memory, planning, tool use and error recovery.
• Develop and fine‑tune small and medium language models (SLMs) for regulated banking use cases, creating custom evaluation frameworks and domain‑specific benchmarks.
• Architect and implement MCP servers, A2A protocols and federated API layers that allow AI agents to operate across the group’s multi‑country infrastructure.
• Prototype new ideas in two‑week sprints: hypothesis, architecture, code, functional demo and one‑pager.
• Collaborate with researchers, data scientists and business stakeholders to translate complex technical concepts into tangible bank value.
• Monitor emerging research, evaluate new tools and integrate them into the lab’s workflow when they add real value.
• Produce clean, tested, observable code ready for production scaling; own code end‑to‑end.

• 4–8 years of software or AI engineering, with at least 2 years building and maintaining LLM‑powered systems in production environments.
• Hands‑on experience designing and deploying multi‑agent AI systems with real‑world complexity: memory management, stateful orchestration, tool use, multi‑step planning and graceful failure recovery.
• Experience fine‑tuning or adapting language models (SFT, LoRA, RLHF) for domain‑specific tasks, including dataset curation and evaluation design.
• Track record of delivering complete systems independently within tight timelines—architecture decision to production‑ready code.
• Experience building and consuming REST APIs and integrating AI systems with enterprise data sources, cloud services and third‑party platforms.
• Prior experience in banking, fintech or other regulated industries (Preferred).
• Exposure to Harness Engineering methodologies (Preferred).

• Bachelor’s or Master’s degree in Computer Science, Software Engineering, Mathematics, Physics or an equivalent technical field.
• Master’s degree or equivalent advanced qualification in AI, Machine Learning or related discipline (Preferred).

• English – professional working proficiency (Required).
• Spanish – professional working proficiency (Required).

• Python – advanced proficiency, clean, tested, production‑grade code.
• Agentic AI frameworks:
  Lang Graph, Auto Gen, CrewAI or Semantic Kernel.
• LLM APIs and model ecosystems:
  Anthropic Claude API, OpenAI, open‑source models (Llama, Mistral). Understand cost, latency and quality trade‑offs.
• Cloud infrastructure: AWS (Bedrock, Lambda, Sage Maker, S3).
• Model evaluation and observability: design evaluations, LLM‑as‑judge, RAGAS, Promptfoo, Langfuse or equivalents.
• API development:
  FastAPI or equivalent.
• Dev Ops basics:
  Docker, Git, CI/CD pipelines.
• MCP server design and implementation (Preferred).
• SLM fine‑tuning pipelines: vLLM, Ollama, Unsloth or equivalents (Preferred).
• Knowledge graphs or GraphRAG for structured knowledge retrieval (Preferred).
• A2A protocol, x402 or AP2 for agentic payments or agent‑to‑agent communication (Preferred).
• Kubernetes, Terraform or equivalents for production‑scale deployment (Preferred).

• Radical autonomy: set priorities independently.
• Builder’s mindset: build proofs of concept rather than just slides.
• Speed with judgment: deliver functional demos in two weeks and know when to transfer or refine.
• Frontier awareness: stay current on research and industry developments.
• Communication across roles: explain complex architectures to stakeholders and write technical one‑pagers.
• Collaborative rigor: give and receive direct technical feedback and document decisions.

• Hybrid working model with flexible hours.
• Access to extensive learning platforms, including Santander Open Academy.
• Frontier exposure: work with Anthropic, Sakana AI, AWS, CMU, ICMAT and INRIA.
• Research impact: publication opportunities on arXiv.
• Competitive…