The "HPC Platform" training program has been launched

We have launched the "HPC Platform" (hpc-study.ru, in Russian) — a training program in high-performance computing, artificial intelligence and data processing. Its main principle is that practice matters more than theory: classes are held not on slides, but on a real training cluster with the Slurm scheduler, a parallel file system, InfiniBand and GPUs.

The training is organized by the non-profit partnership "HPC Platform" — an open community of engineers, researchers and students. The partnership runs professional development programs, open seminars and thematic groups and chats where participants help each other.

Professional development programs

• HPC Systems Engineer (intermediate level, 96 hours, 12 weeks) — from the physical design of a cluster to monitoring and organizing parallel computations: Slurm, MPI and OpenMP, CUDA and NCCL, Lustre and BeeGFS, Prometheus and Grafana. The program ends with a final project: students deploy a training cluster, set up monitoring and alerting, and run and optimize a parallel job. The next cohort starts on November 6, 2026 in a hybrid format, and the following one starts online on February 6, 2027.
• Computing Environment Administrator (basic level, 150 hours) — networks and network services, storage and file systems, on-demand cloud environments, multi-user HPC and infrastructure monitoring. The online evening cohort starts on November 20, 2026.
• DevOps / MLOps Engineer (intermediate level, 150 hours) — automating the delivery of code and models: GitLab CI/CD, MLflow and Optuna, Apache Airflow, Prometheus and Grafana, including in Kubernetes. The in-person intensive course starts on December 4, 2026.
• NVIDIA DGX SuperPOD Infrastructure (advanced level, 160 hours) — operating an AI supercomputer: DGX A100/H100 servers, InfiniBand networks, the NVIDIA software stack, Slurm, monitoring and troubleshooting. The program is held in person for corporate groups, with the next start on December 18, 2026.

Each cohort has a private chat with curators. The community also hosts open interest groups: the partnership's training cluster, InfiniBand and RoCE networks, GPUs and distributed training, Slurm and schedulers.

How to get started

1. Submit an application on the page of the chosen program and tell about your experience with Linux and networks.
2. Have a short interview — a curator will contact you within two business days and help you choose a program and an entry level.
3. On the cohort's start day, get an account on the training cluster and access to the group chat.

Details about the programs and the schedule of cohort starts and open seminars are available at hpc-study.ru.