Compute Infrastructure Engineer – NeuraGym (human)

The role

You will design, build, and operate a scalable compute and storage infrastructure for multimodal robot data. You will turn servers, workstations, and edge devices into a distributed compute system to enable embodied AI systems.

What you will do

• Plan and design NeuraGym server infrastructure including compute, GPU, storage, network, rack layout, power, and cooling.
• Build and operate virtualisation environments based on hypervisors.
• Design and operate Kubernetes clusters for data preprocessing workloads.
• Distribute compute- and data-intensive processes across multiple machines.
• Design network infrastructure including VLANs, switching, bandwidth planning, and firewall rules.
• Build storage and data pipelines for high data volumes.
• Automate provisioning, deployment, and operations with infrastructure-as-code and CI/CD tooling.
• Identify and resolve performance, stability, and security issues.
• Develop backup, recovery, and access concepts.

Required qualifications

• Degree in computer science, systems engineering, engineering or a related field, or comparable vocational training with relevant practical experience.
• 4+ years in system administration, infrastructure engineering, DevOps or platform engineering.
• Solid experience planning, designing and operating server infrastructure.
• Hands-on experience with hypervisors and virtualisation (e.g. Proxmox, KVM, VMware).
• Strong knowledge of container technologies and orchestration (Docker, Kubernetes or comparable platforms).
• Understanding of how processes, data and compute load are distributed, synchronised and monitored across multiple machines.
• Confident with Linux systems, shell scripting and at least one scripting language (e.g. Python, Bash).
• Experience with infrastructure as code and automation (e.g. Ansible, Terraform) as well as CI/CD tooling (e.g. GitLab CI).
• A structured, self-driven way of working.

Nice to have

• Experience with monitoring and logging stacks (e.g. Prometheus, Grafana)
• Ideally with GPU workloads

About the company

neura-robotics-gmbh