At Avalon Solar, I evaluated 30 AI models against defined metrics, documenting accuracy, runtime, and cost trade-offs to inform model selection. I also re-architected the EL image-stitching pipeline in C++ with OpenCV and CUDA, improving output quality by 60%.
I built and maintained an internal full-stack platform for AI training, testing, and pre-deployment quality control, giving the team a repeatable QC gate before release.
At Cegedim SRH, I built and maintained internal Python/Django web platforms, delivering features for a new application and later extending the platform used daily by the IT support team. At CPAM, I took an internal PHP MVC application for Linux CRON task management from concept to production.
On PharmaXcess, I owned the backend across a four-person team and built a Raspberry Pi proof of concept for an automated medicine dispenser. I implemented its prescription OCR pipeline and deployed the backend with Docker, CI/CD, and automated tests. I’m also developing Genesis, a governance platform for AI coding agents.
