In June 2026 I finished both degrees: my Master’s and Ingénieur d’État in AI & Data Science at ENSIA and my Master’s in Management for Engineers at EHEC. In September 2026 I started a PhD in Computer Science at Université de Montréal (DIRO).
What I’m working on
My research is about AI-assisted software engineering, and specifically about AI coding agents: tools that read a codebase, plan, edit files, and run tests with little human input.
They’re getting adopted quickly, but we know surprisingly little about them as engineering tools. The questions I care about are:
- Reliability. When an agent says it fixed something, how often is it right, and how do we tell?
- Cost. How many tokens, how much time, and how much money does a given result take? Where is that budget wasted?
- Open vs. proprietary. How do open-source agents and models compare with proprietary ones, task for task, on fair benchmarks?
These questions come straight from my internship. The context-retrieval work was about cost. The LLM-as-Jury was about reliability. Comparing toolchains is the obvious next step.
Looking back
Looking back, the story so far fits together. I started by building products for real users. I learned that a product is only as good as people’s trust in it. Now I’m studying the tools that are changing how software is built, to measure when they can be trusted.
I’ll keep writing here as papers come out. Each one gets its own post in the Papers section of the blog.