Join the Swiss National Supercomputing Centre (CSCS)
The Swiss National Supercomputing Centre (CSCS) is at the forefront of high-performance computing and data research, providing essential support for world-class science in Switzerland. Operated by ETH Zurich, we welcome researchers from academia, industry, and the business sector at our facilities in Lugano and Zurich.
Position Overview
We are seeking a candidate for a two-year position, with the work location being either Lugano or Zürich. This role is critical in bridging the gap between raw data storage and providing researchers with reliable, accessible, and usable datasets.
Responsibilities
- Bridge ingestion and use: Design processes and metadata that transform ingested data into easily findable and consumable formats, such as catalogs and schemas tailored to training jobs and simulations.
- Ensure data traceability: Develop lineage and provenance to ensure any dataset can be traced back to its origins and transformations, prioritizing reproducibility as a core requirement.
- Tune for workload optimization: Optimize parallel filesystems (Lustre, GPFS) and object storage to cater to the concurrency and patterns of distributed GPU training and HPC simulations.
- Operate at scale safely: Design and manage multi-petabyte storage systems ensuring data integrity and availability through techniques like erasure coding and hot-to-archival tiering.
- Automate processes: Implement storage and data services as code for efficient scaling at large volumes.
- Make operations observable: Monitor storage health and performance metrics proactively to identify potential issues before they impact users.
- Translate technical requirements: Convert the real access patterns reported by scientists and ML engineers into actionable technical requirements.
Qualifications
- A technical degree in Computer Science, Engineering, or equivalent experience.
- Solid grounding in storage systems, including filesystems and performance tuning.
- Proficiency in Python and experience in automating infrastructure (e.g., Ansible, Terraform).
- Understanding of how ML and scientific workloads interact with data.
- A perspective on data lineage, provenance, or reproducibility, with relevant tooling experience.
Preferred Skills
- Experience with parallel filesystems (Lustre, Spectrum Scale/GPFS) or distributed storage systems (Ceph, VAST).
- Familiarity with scientific data formats (HDF5, Zarr, Parquet) and their appropriate use cases.
- Experience with object storage (S3) in conjunction with ML frameworks (PyTorch, TensorFlow).
- Knowledge of orchestration tools (Kubernetes, Argo) and data movement strategies.
- Experience with data versioning and cataloging (e.g., DVC, lakeFS) and an understanding of FAIR data principles.
What We Offer
At CSCS, you will encounter hardware and challenges that surpass typical enterprise IT environments, directly contributing to published science and advanced model training. You'll have the opportunity to shape data management practices and expand your skill set in a collaborative and innovative environment.
Core Values
- Curiosity: A passion for learning and understanding systems deeply.
- Openness: Collaboration and valuing diverse perspectives.
- Courage: Willingness to tackle challenging and unfamiliar problems.
- Supportive: A commitment to helping colleagues and users succeed.
- Integrity: Acting responsibly, reliably, and transparently.
Encouragement for Diverse Applicants
We are dedicated to building a diverse and inclusive engineering team and strongly encourage applications from underrepresented groups in technology. Your impact will extend beyond your role—join ETH Zurich in contributing to meaningful societal change.
Benefits
CSCS offers numerous benefits, including public transport season tickets, a variety of sports activities, childcare support, and attractive pension options, creating a rewarding and balanced work environment.
How to Apply
If you are interested in this opportunity, apply online using the form below. Please note that only applications matching the job profile will be considered.
Further Information
For more about CSCS, please visit our website. Any questions regarding the position can be directed to Pim Witlox at witlox@cscs.ch (no applications).