Orchestration Workload Engineer – ACE – AI Factory

About the role

As a Workload Orchestration Engineer in the Accelerated Compute Engineering (ACE) team, you will be the internal expert for workload orchestration, owning and advancing the SLURM scheduler stack across Roche’s high‑performance computing (HPC) platforms. You will bridge traditional scientific computing with modern AI workloads and mentor colleagues to develop technical expertise.

Key responsibilities

• Architect, scale and maintain SLURM across heterogeneous HPC and AI environments, ensuring high availability and dynamic resource distribution.
• Design advanced SLURM configurations, including custom plugins, topology‑aware scheduling, GPU management, priority trees and QoS policies.
• Integrate SLURM with Kubernetes and other orchestration platforms to streamline job submission across on‑prem and cloud architectures.
• Implement container standards (Singularity/Apptainer) within SLURM and support hybrid AI/HPC workloads.
• Lead global cross‑functional initiatives to establish orchestration standards, policies and architectural patterns.
• Partner with observability engineers to build telemetry dashboards for job efficiency, queue times and hardware utilization.

Required profile

• Bachelor’s or advanced degree in Computer Science, Applied Mathematics, Computational Engineering or related field.
• Extensive systems engineering experience with deep specialization in workload scheduling, SLURM administration and multi‑tenant cluster optimization.
• Proven track record of leading complex technical initiatives and mentoring engineering peers.
• Experience in life‑sciences, pharmaceutical R&D or high‑performance scientific research environments.

Required skills

• SLURM architecture, scaling, tuning, plugin integration and GRES/TRES modeling.
• Kubernetes fundamentals and container runtimes (Singularity, Apptainer, Docker, Enroot).
• GPU scheduling technologies (NVIDIA MIG, fractionalization) and high‑speed interconnects (InfiniBand, RoCE).
• MPI, NCCL and multi‑node communication frameworks.
• Infrastructure‑as‑Code tools such as Ansible and Terraform.
• Observability and accounting tools (SlurmDBD, telemetry dashboards).