An online technical session on making deep-learning training and inference faster on GPUs, covering CUDA fundamentals and practical PyTorch performance-optimisation techniques: memory layout, kernel fusion, mixed precision and profiling workflows. Aimed at ML engineers and researchers who train or serve models and want to reduce cost and latency rather than add more compute. Hosted by the Dubai-based AI Performance Engineering Meetup community for its members and the wider UAE ML-engineering audience. Free, online, RSVP required.

From the Mon 19 Oct listing on Meetup. Each date's page has its own details.