Modern computational science and AI/ML research demand flexible, reproducible software environments to support the development of robust and reliable workflows. However, the vast majority of researchers manage these environments interactively, leading to fragile, non-portable setups that are difficult to share. This talk will cover best practices for managing software environments using declarative and stateless principles from cloud-native ecosystems - adapted for batch-based HPC systems. By adopting this approach, researchers can minimize "works on my machine" issues, accelerate peer adoption, and streamline the transition from prototyping to large-scale HPC/AI workloads. 

This session introduces the GPU computing services offered by UW’s Research Computing team, featuring high-performance hardware and a robust support ecosystem. We provide tailored resources for AI, machine learning, and scientific computing, along with personalized consultations, training, and domain-specific expertise. Whether you're new to GPU computing or scaling advanced research, we’re here to help you succeed.

NVIDIA RAPIDS is an open-source suite of libraries designed to accelerate data science and machine learning workflows on GPUs. Build on CUDA but designed with a user-friendly Python front-end, RAPIDS enables end-to-end data pipelines from data loading and preprocessing to model training entirely on GPUs, drastically reducing computation time compared to CPU-based workflows. This presentation will showcase how NVIDIA RAPIDS can serve as a drop-in replacement for popular Python libraries like pandas, scikit-learn, and networkx by running entirely on GPUs and accelerating your workflows.

As GPUs revolutionize computational science and AI/ML research, integrating GPU computing into curricula has become essential for preparing students to tackle modern scientific and engineering challenges. This talk explores strategies for embedding GPU computing concepts, tools, and workflows into graduate and undergraduate courses, focusing on hands-on learning and practical applications in scientific simulations and AI/ML training pipelines.

Post-event opportunity to ask the Cambridge, NVIDIA, and UW-IT Research Computing team any and all questions from today's presentations