Built and trained a Neural Network with pyTorch to generate robust invisible forensic watermarks on a GPU with 5x real-time transcoding.
H.264, H.265, J2K codec optimization, custom video and audio filters integration in FFmpeg.
Origin and Edge JavaScript cloud video secure processing for Live and VOD (Encrypted/signed tokens & custom manifests)
Fully functionnal Netflix clone created from scratch, with NodeJS+SQL backend and ABR HLS streaming.
AWS auto-scaling video transcoding backend, with automatic quality control, Job management with AWS RDS databases.
MATLAB, Python, C and C++ signal processing for Fingerprinting, Watermarking, Feature Extraction, Face Detection, ...
FPGA and DSP electronic design and validation for Embedded Systems.
Video editing C++/Qt desktop applications, Docker containers with cross-compilation, Linux drivers, Anti-reverse engineering, Secure license management, and more...
Projects usually start with algorithm validation and optimization in MATLAB or Python before being ported to C or C++. Robustness, modularity and code readability are a priority. By combining software engineering, traditionnal signal processing, machine learning, cloud processing and cybersecurity practices, we work to develop modern solutions that solve your problems in video processing and data protection, whether your system is a large cloud deployment, a tiny embedded system or a PC application.