Research Engineer

  • Amsterdam, Netherlands
  • Full-Time
  • On-Site

Job Description:

Research Engineer (AI / LLM / Agents)
Amsterdam, Netherlands; Belgrade, Serbia; Limassol, Cyprus; London, United Kingdom; Munich, Germany; Remote

We are currently looking for a Research Engineer for one of our clients working on an innovative AI product in the creator economy space. The project focuses on building a next-generation platform that helps creators interact with their audiences in new ways, transform ideas into content, and build sustainable businesses using LLMs and AI agents.

This role is ideal for someone with a strong background in engineering or machine learning who enjoys turning cutting-edge ideas into real, shipped products.

Our client operates in a fast-paced, experimentation-driven environment. Teams iterate quickly, ship early versions, measure results, and continuously improve products based on data. Delivery, experimentation, and ownership are core parts of the culture.

In this role you will:

  • Ship AI product features that combine engineering and machine learning.

  • Build and improve agent workflows (tooling, retries, guardrails, fallbacks).

  • Improve the latency, cost, and reliability of AI systems.

  • Stay up to date with developments in the AI ecosystem (agent frameworks, tooling, emerging methods) and turn them into production-ready features.

  • Build self-improving systems that run experiments, learn from feedback and evaluations, and continuously improve the product.

We are looking for someone who:

  • Has experience shipping production systems, not only prototypes or demos.

  • Has worked with LLMs in real-world applications (structured outputs, function calling, tool usage, reliability improvements).

  • Has experience building AI agents (multi-step workflows that call tools, maintain state, and handle failures).

  • Has implemented evaluation frameworks to track quality (tests, regression checks, evaluation harnesses).

  • Is comfortable working in a fast-moving environment, iterating quickly, and owning outcomes end-to-end.