AI Engineer

We’re looking for an engineer to build and run agentic workloads that operate unattended, from data extraction through to orchestration interfaces. You’ll build the data pipelines that ingest, extract and transform a large enterprise document corpus into something an LLM can actually retrieve against, and the sandboxed agents that work over it every night without a human watching.

You should enjoy the hard parts: hybrid retrieval at scale, data extraction that survives messy real-world documents, and exposing sensitive data to agents in a controlled, permission-aware way.

• Experience integrating software components into a production-grade platform for enterprise use.
• Experience exposing sensitive data to agentic workloads / LLM harnesses in a controlled manner.

• Hands-on search and retrieval experience: chunking, hybrid keyword and vector retrieval, reranking, and measuring whether the results are actually any good.

• Strong modern software engineering skills: comfortable with Linux, bash, git, containerisation (Docker or similar), and authn / authz / access-control paradigms.

• A clear grasp of the limits of “vibe coding” in a team or enterprise setting, and of when code needs to be understandable, extensible and transferable.

• Experience building data transformation and extraction pipelines. Nice to have

• In-depth Python and Go knowledge.
• In-depth Kubernetes knowledge.
• Experience with LiteLLM.
• Experience setting up internal chat or agent platforms (OpenWebUI, ComfyUI, LibreChat or similar).
• Experience with open-source agent harnesses.
• Demonstrable skill in manually creating, or ideally auto-optimising, agent skills and agentic workloads.
• Experience with low-code orchestration tools such as n8n.
• OCR at volume (Azure Document Intelligence or similar).

• Strong PostgreSQL skills: schema design, query performance at scale, and tuning pgvector or comparable vector-search tools. Not the right fit if you’re

• A pure data scientist, focused on explaining patterns in data.
• A pure academic, chasing architectural or conceptual purity.
• A pure coder, looking to build or own a single application.

• A demo engineer. We’re building a system that has to run unattended, every night. How you work

• You reliably execute a plan.
• You communicate well and are comfortable talking to people.
• You take ownership of parts of a project without being controlling.
• You simplify or delete code rather than pile it on.
• You raise blockers with the team rather than working around them in isolation.

About Gapstars At Gapstars, we partner with some of Europe’s most ambitious companies, from disruptive tech startups to established Benelux accounting firms, helping them build high-performing teams in engineering and finance.

Headquartered in the Netherlands, with talent hubs in Sri Lanka and Portugal, we are home to 300+ professionals who thrive on solving real-world challenges with modern solutions. Our teams work across domains, from networking, marketplaces, SaaS, and AI to embedded finance and accounting, delivering scalable solutions that drive meaningful outcomes., * Data pipelines that ingest, extract and transform a large enterprise document corpus into an LLM-retrievable form.

• Hybrid retrieval over that corpus: chunking, keyword plus vector search, reranking, and measurement of answer quality.
• Sandboxed agents that run scheduled, unattended workloads over the indexed data.
• Permission-aware access so agents only ever reach data the requesting user is entitled to see.