Senior Director, Developer Technologies (Deep Learning)
CurrentAt NVIDIA, Andrew directs technical teams working on a variety of projects spanning large cross-company software launches to focused deep learning R&D experiments. The most recently announced project is "NVIDIA DLSS Frame Generation" which is revolutionary neural graphics technology launched as part of NVIDIA's latest Ada Lovelace GPU.Using situational leadership with direct reports and "matrix teams" of up to 50, Andrew mixes management between Agile (JIRA) and more flexible approaches for longer-term R&D. Software typically starts in Python using a mix of the popular deep learning frameworks such as PyTorch, JAX and TensorFlow. For production, the software is moved to C++, TensorRT, CUDA, CuDNN and CuBLAS and is optimized for GPU hardware both for high performance training and real-time inference (targets are often well below 5ms for full model evaluation).Andrew has strong communication skills and deals well with:- Technical colleagues & subject matter experts.- Product & marketing teams.- NVIDIA executives.- Industry experts (with recent presentations at GDC & GTC conferences).HIGHLIGHTED PROJECT:NVIDIA DLSS 3 Frame Generation- Next generation real-time neural graphics using exciting new deep learning model to drastically boost 3D performance.- Leverages NVIDIA Ada Lovelace GPU features and delivered on time for product launch.- Extensive optimizations for maximum CUDA and NVIDIA TensorCore throughput.- Over 30 games and apps signed on at launch.- Expanded from bespoke library to SDK with detailed documentation, debugging tools and sample code.- Took technology from R&D to product design, testing and launch; led team of 15-20 specialist engineers.- Deep involvement with technical review staff, product marketing and executive decision makers.