Optimizing the Performance of Machine Learning in Enterprise Java SaaS with GraalVM, Python and CUDA

What if you could easily integrate your favorite Python ML library into your enterprise Java application, or even take advantage of GPU acceleration for model training, while barely changing your application stack? We'll use GraalVM to add recommender system algorithms written in Python to Netsuite SuiteCommerce. Additionally, we show how the new GraalVM CUDA connector can be used to quickly offload work to GPUs and interact with the CUDA runtime. Lukas Stadler is a researcher at Oracle Labs Austria and manages the Python, LLVM and R runtimes of the GraalVM polyglot language ecosystem. He works on all aspects of the Graal and Truffle projects, with a special focus on novel compiler optimizations and dynamic language implementations. His research interests include novel techniques for programming language execution, Virtual Machines, dynamic compiler optimizations and feedback-directed optimizations. Radek is a software engineer at Oracle NetSuite. He is a machine learning and artificial intelligence enthusiast, currently focused on designing and productionisation of the next generation product recommender system for SuiteCommerce.