Forward Deployed AI Engineer

Excerpt from the original posting

About RevoData 
RevoData is a boutique Data & AI consultancy and proud Databricks Gold partner. We focus exclusively on Databricks, allowing us to go deep and do things properly. We help companies across Europe unlock the value in their data, helping them design and build scalable platforms that support real business use cases. 
Having built our foundations in the Netherlands, we’re now expanding beyond our home market and growing our team in Hungary. 
We foster a no-ego, no-nonsense culture where collaboration, proactivity and knowledge sharing are part of everyday work. 
Sound like your kind of environment? Then keep scrolling! 

You are a builder who thrives as much in the first as in the last mile of AI - the space between a cool demo and a robust, production-ready system. You enjoy turning AI capabilities into real-world applications that people actually use, and you don't just ship a working prototype; you can prove it works, then make it measurably better.
As a Forward Deployed AI Engineer at RevoData, you embed with clients to design and build AI solutions that solve real business problems. You combine strong technical skills with curiosity and pragmatism, and you enjoy bridging the gap between experimentation and production.
We'd love to talk if: 
- You've built AI systems that run in the real world: You have experience across the full lifecycle - from early experimentation to deployment and monitoring - and understand what it takes to make AI systems reliable in production.
- You're experienced with LLMs & GenAI: You've moved beyond simple prompts and have built production-grade GenAI applications, including RAG architectures, agentic workflows, and fine-tuning, using platforms like OpenAI, Anthropic, Azure AI, or similar.
- You're comfortable with GenAI frameworks: You've used tools like LangChain, LangGraph, Pydantic AI, Hugging Face, or DSPy, and understand how to structure and orchestrate AI-driven applications beyond simple prototypes.
- You're evaluation-driven: You don't trust "it seems better." You establish a baseline, define metrics that actually capture the failure modes that matter, and iterate against them - so you can show with numbers that a change to a prompt, retrieval step, model, or parameter genuinely moved the needle.
- You understand LLMOps & MLOps: You understand that AI is only as good as its lifecycle management. You're comfortable with experiment tracking, model versioning, observability, and deployment (e.g. MLflow or c…

→ RevoData · Workable