DevOps Engineer for HEP GPU Benchmarks

About the role

The Worldwide LHC Computing Grid (WLCG) is a global collaboration of around 170 computing centres supporting the High-Luminosity LHC era, with growing demand and heterogeneous hardware (CPU, GPU, cloud, HPC). The HEP community uses benchmarking tools to measure and compare computing resources. As a Graduate / Applied Engineer (QUEST) in CERN's IT Department, you will support the development, operation and user support of WLCG HEPScore benchmarking services and help build the next-generation benchmarking software stack.

Key responsibilities

• Develop, package and distribute the HEP benchmark software stack, integrating, profiling and validating GPU workloads from LHC experiment software (event generation, simulation, fast inference, HLT)
• Build, test and maintain container images of HEP workloads via automated GitLab CI/CD pipelines
• Maintain the benchmarking infrastructure and data ingestion services using Puppet, GitLab CI/CD, ActiveMQ, Logstash, OpenSearch, Kibana, Apache Spark and Grafana
• Analyse grid-wide benchmark data to evaluate performance scaling across CPU/GPU architectures and server configurations
• Provide technical support and documentation to the WLCG user community

Required profile

• Nationality of a CERN Member State or Associate Member State (currently Pakistani, Lithuanian, Latvian, Cypriot, Croatian and Turkish applicants are excluded)
• Master's degree with 2 to 6 years of professional experience after graduation, or a PhD with a maximum of 3 years; bachelor's degree holders are not eligible
• No previous CERN fellow or graduate contract
• Studies focused on Applied Physics or Software Engineering
• Spoken and written English, with a commitment to learn French

Required skills

• Python, including modular code, packaging, distribution and testing
• Docker and Apptainer/Singularity, with GPU container runtimes
• GitLab CI/CD, Puppet, Linux administration and shell scripting
• OpenSearch/Elasticsearch, Logstash, Kibana, Grafana, Apache Spark, ActiveMQ
• CUDA, NVIDIA Container Toolkit, AMD ROCm and GPU profiling

What we offer

• Monthly net stipend of 6372 or 7004 Swiss Francs depending on degree and experience
• 30 days paid leave plus 2 weeks annual closure
• Health insurance coverage and pension fund membership
• Relocation package and family allowances
• Contract extendable up to 36 months, with on-the-job and formal training