Hi, I'm Kacem Mathlouthi

I'm a software engineer at Callab AI (YC P26). I joined before Y Combinator and went through the batch with the team, and I build an AI voice agents platform for on-prem telephony that works with existing PBX, SBC, and contact center infrastructure, no migration needed.

Before that, I spent the summer as a Google Summer of Code '25 fellow at NRNB, where I built VCell-AI, an AI agent platform that lets computational biology researchers query, explore, and generate biomodels in natural language.

Earlier, I was a machine learning engineering intern at Orange, building an edge-optimized CNN for plant disease classification and an agentic report generation pipeline.

Outside of work, I build things I wish existed. Right now I'm building Animus, a platform that turns any topic into narrated, research-grounded Manim explainer videos. Before that I built Metis, an AI platform for GitHub pull requests with a cloud coding agent for autonomous code review, issue resolution, and PR summaries.

On the research side, I have a paper accepted at TMLR, titled “Generalization Measures under Controlled Covariate Shift: A Regime-Aware Benchmark”, which asks whether generalization measures that look reliable under IID evaluation still hold up when image classifiers face corruptions and perturbations, and finds that their usefulness is strongly regime-dependent.

I'm currently a software engineering student at INSAT in Tunis, graduating in 2027, where I also served as Technical Manager for AINS 3.0, the AI National Summit by IEEE CS INSAT, leading a three-track hackathon, a data science competition, and four workshops. I like building things that put AI to practical use.