SUMMARY
HPC Expert (ClearML & Confidential Computing)
Summary
This position is dedicated to an HPC Expert specializing in ClearML infrastructure and Confidential Computing technologies. The primary objective is to oversee the deployment, management, and security of high-performance computing environments that support advanced AI and machine learning initiatives. The role is critical for ensuring the reliability, scalability, and security of data pipelines, model registries, and computational resources, enabling seamless execution of complex workloads across diverse hardware architectures.
Location: Bangalore
Experience: 8-13 years
Responsibilities
- Oversee the administration of ClearML Server, including the configuration of execution queues, user management, role-based access control (RBAC), experiment tracking, data pipelines, datasets, artifacts, and the model registry.
- Deploy and tune ClearML Agents on CPU and GPU worker nodes, ensuring seamless integration with major HPC scheduling systems like Slurm, PBS Professional, LSF, or Kubernetes.
- Facilitate the execution of GPU-accelerated AI and ML workloads by managing NVIDIA drivers, CUDA libraries, NCCL, UCX, and containerized application environments.
- Enforce secure container execution protocols utilizing tools such as Apptainer/Singularity, Docker, Enroot, Pyxis, or Kubernetes.
- Deploy and maintain confidential computing solutions, including AMD SEV-SNP, Intel TDX, NVIDIA Confidential Computing, Secure Boot, TPM, and remote attestation mechanisms.
- Establish robust security frameworks covering authentication, RBAC, IAM, TLS certificate management, secrets handling, and comprehensive audit logging.
- Continuously monitor the health of ClearML services, agent performance, queue status, GPU utilization, task completion rates, and scheduler connectivity.
- Drive automation for deployment, configuration, system monitoring, and troubleshooting through the use of Python, Bash, Ansible, and Git.
Requirements
- Proven expertise in Linux system administration, specifically with distributions such as RHEL, SLES, Rocky Linux, or Ubuntu.
- Demonstrated hands-on experience administering ClearML or equivalent MLOps platforms.
- Familiarity with HPC workload schedulers including Slurm, PBS Professional, or LSF.
- Solid comprehension of NVIDIA GPU architecture, driver management, CUDA, GPU performance monitoring, and the principles of distributed training.
- Proficiency with containerization technologies like Apptainer/Singularity, Docker, Enroot, Pyxis, or Kubernetes.
- Advanced scripting and automation capabilities using Python, Bash, Ansible, and Git.
- Strong foundation in security principles, encompassing RBAC, IAM, TLS, certificate management, secrets storage, Secure Boot, and audit trails.
- Knowledge of confidential computing paradigms, including Trusted Execution Environments (TEE), memory encryption, TPM, remote attestation, AMD SEV-SNP, and Intel TDX.
