Artificial Intelligence Engineer

About Duely

Duely is an AI-native legal service provider for M\&A. Most "legal AI" sells tools to lawyers and lets them keep billing by the hour. We don't. We do the legal M\&A work ourselves - data room reviews, gap analyses, contract summaries, full DD reports - powered by our own AI engine and verified by our legal experts. We deliver the report, we take responsibility for the output, and we charge per page, not per billable hour.

We don't sell to law firms. We compete with them, and sell to their clients - the PE funds, family offices, and corporate M\&A teams who are tired of six-figure bills and weeks of waiting. The incumbents won't follow us here: no law firm will cannibalize its own billable hours. That's the opening, and it won't stay open forever.

We're already moving: first AI-native provider in the Benelux to raise venture capital in this space (€1M\+), with our engine running live on a global virtual data room platform that sees \~1,800 M\&A deals a year. Every deal makes the engine better - and our engineers iterate on real deals, daily, with a legal expert sitting right next to them. We're a small, on-site team in Ghent, building something rare and looking for people to join us!

The Role

As an AI Engineer at Duely, you'll work on the engines that turn messy data rooms - thousands of legal and financial documents - into trusted deliverables. This is hands-on LLM application engineering, not model training. Your work will include:

• Designing and shipping LLM-powered features (extraction, summarization, Q\&A, document analysis, and agentic workflows) into our production platform
• Building rigorous evaluation pipelines - measuring quality is the core of the job, because lawyers depend on the output
• Improving accuracy and reliability through prompting, retrieval, and context/harness engineering
• Designing the infrastructure and APIs that make this fast, reliable, and cost-efficient at the scale of thousands of documents per deal - unlocking search and reasoning across a data room that used to take a team of associates weeks
• Working closely with legal and product teams to turn real M\&A workflows into AI systems that hold up under scrutiny

This is a full-time, on-site role in Ghent.

What we're looking for

• Strong software engineering fundamentals and fluency in Python (production code, not just notebooks)
• Hands-on experience building with LLMs - prompting, RAG, evals, agents, context engineering, or similar (whether at work or on serious side projects)
• A bias toward shipping and measuring, and the judgment to know when an output is actually good enough to put in front of a client
• Clear communication - you can explain a technical tradeoff to a lawyer and a legal nuance to an engineer
• Bonus: exposure to legal tech, M\&A, or other high-stakes / document-heavy domains; experience with observability, data privacy, or cloud (we run on Azure)
• Our stack: Python throughout, Azure OpenAI for the LLM layer, PostgreSQL for storage, Flask APIs, and async background processing - all on Azure. Strong observability and evals (OpenTel) sit at the core, because in this domain you can't improve what you can't measure.

What you get

• Real equity. You're still early - you own a meaningful piece of what we build, not a rounding error.
• Ownership from day one - you'll ship features that clients use, not tickets in a backlog.
• A lovely, small team where your work is visible and your judgment is trusted.
• A rare shot at defining how an entire profession adopts AI, before everyone else catches on.
• Ah yes, and a competitive salary.

We care more about your drive, what you've built and how you think than about your degree or years of experience.