On 28 th November, ISSAI researchers participated in the High-Performance Computing (HPC) and Artificial Intelligence (AI) Symposium, organized by Nazarbayev University (NU) School of Sciences and Humanities professors Dr. Ernazar Abdikamalov, Dr. Sergey Bubin, Dr. Michael Lewis, and Dr. Bekdaulet Shukirgaliyev. The aim of the event was to raise awareness of current projects and computational resources, promote interaction, and collaboration between different research groups, and get a better picture of NU’s computational needs to drive the ongoing development of HPC & AI resources at the University.
During the symposium, different research groups and professors from different disciplines presented the computational aspects of their work and how they use these technologies to drive their disciplines forward. ISSAI members spoke about what computing resources they have and how they use them for their projects. For instance, ISSAI technical project coordinator Yerbol Absalyamov showed ISSAI’s resources, such as HPC_AMD, NVIDIA DGX 1, 2, and A100 and mentioned their advantages and disadvantages. In addition, Yerbol mentioned that ISSAI shares its computational resources with several NU departments and other universities in Kazakhstan.
ISSAI data scientists Madina Abdrakhmanova and Saida Mussakhojayeva spoke about their experience of using HPC. For instance, Madina gave a presentation on the analysis of computational resources with the growth of training datasets and her experience of using computational resources for the Speaking Faces project, and Saida spoke about resource constraints in speech processing research. They explained how AI computational infrastructure helps in processing data, but also the requirements and constraints they face while working on their projects. In addition, computer engineer Shakh-Izat Nurgaliyev presented The HPC for Small and Tiny: TinyML-based Wearable Tech for the Visually Impaired and Blind. He explained how NVIDIA infrastructures can be used for the machine learning model and deployment pipeline of the project.