Machine Learning: - Research and Develop ML solutions for 3D reconstruction (Gaussian Splatting approaches) - Optimize and deploy deep learning feature extraction and matching models (SuperPoint, LightGlue) on embedded Linux hardware using TensorRT - Quantize PyTorch feature extraction models to FP16 and INT8 via PTQ, implementing zero-copy C++ CUDA memory buffers to eliminate host-to-device bottlenecks - Profile neural network execution and telemetry streaming using NVIDIA Nsight Systems and PyTorch Profiler Robotics: - Development, deployment, and maintenance of high-performance distributed VIO and SLAM system architectures in C++ and Python - Design and implement autonomy (perception, navigation) stack based on ROS2/Nav2 and in-house distributed system based on Zenoh - Develop a scalable data visualization software backend (point clouds, trajectories, telemetry...) - Performance monitoring and debugging with the Grafana stack - Agile development and cross-team communication and coordination - Lead an engineering team to deliver functional software systems

Development of AI-based character locomotion Optimization and deployment of real-time high-fidelity interactive XR software on the edge Bridge the gap between cutting-edge motion research and mobile hardware