GPU, CUDA, and PyTorch Performance Optimizations

Explore GPU, CUDA and PyTorch optimisation for faster AI inference at this online technical meetup on 19 October 2026 at 5:00 PM. Hosted by Chris Fregly for the AI Performance Engineering Meetup (London–Cambridge), this virtual session focuses on improving performance across GPU-powered AI systems. It offers a dedicated space to examine GPU execution, CUDA and PyTorch optimisation through technical talks and group discussion. ## What to expect The programme opens with introductions and meetup updates from Chris Fregly and Antje Barth. That is followed by two talks under the theme GPU, PyTorch and CUDA Performance Optimizations. The published agenda does not name the speakers for these two sessions, but both centre on the event’s core performance-engineering topic. The online format makes this a useful opportunity to: - Explore approaches to GPU and CUDA performance optimisation - Consider how PyTorch workloads can be made more efficient - Connect these techniques to the demands of AI inference - Hear meetup updates and take part in technical discussion - Discover resources for continuing your performance-engineering work This event is particularly relevant to AI and machine-learning engineers, performance engineers, software developers and technical practitioners working with GPU-backed systems. Because the subject matter is specialised, attendees with some familiarity with AI infrastructure, PyTorch, CUDA or GPU computing are likely to gain the most from the talks. The listing does not state formal prerequisites, however, and the introductory opening provides context for people new to the meetup community. ## Format and resources The meetup takes place online via Zoom and forms part of a series scheduled for every third Monday through the end of 2026. Alongside the live programme, the organisers provide several related resources:… Register for this event