DAI.MOE

I am Dai (Arts) Yang , senior solutions architect for DGX, datacenter, and cloud.

I lead technically the solution architecture for DGX Cloud in Europe, Middle East and Africa, at NVIDIA Corp. I am the responsible solution architect for NVIDIA Cambridge-1 previously. My work contact is daiy (a t) nvidia.com.

Connect with me at Linkedin: https://www.linkedin.com/in/dai-yang-de/.

Responsible Projects

• NVIDIA Cambridge-1, (Top500).
• NVIDIA DGX Cloud
• Tokyo-1
• NEOM SuperPOD
• undisclosed Sweden SuperPOD

Articles

NVIDIA DevBlog Author Page

• Autoscaling NVIDIA Riva Deployment with Kubernetes for Speech AI in Production
• Accelerating Volkswagen Connected Car Data Pipelines 100x Faster with NVIDIA RAPIDS
• Getting Kubernetes Ready for the NVIDIA A100 GPU with Multi-Instance GPU
• A complete guide for CPU/single GPU/multi-GPU/multi-Node training using TensorFlow 2.0 + Horovod

Teaching

I teach various guest lectures and partical courses at TUM.

CV

• 2020 - now: Senior Solution Architect Datacenter, NVIDIA
• 2017 - 2019: Research Associate and PhD Student at Chair of Computer Architecture and Parallel Systems, Technical University of Munich.
• 2015 - 2017: Software and Systems Entineer, Airbus Defence and Space.
• 2013 - 2015: IT Infrastructure, UnternehmerTUM gGmbH.

Education

Certifications and Languages

• 2020 NVIDIA Certified DLI Instructor, Fundamentals of CUDA C/C++
• 2019 Certified Scrum Master
• Full professional+: German, Mandarin Chinese, English
• Hobbiest: Japanese

Degrees

• 2020 Dr. rer. nat. in Informatics from TU Munich
• 2017 M.Sc. in Informatics from TU Munich
• 2015 B.Sc. in Informatics from TU Munich

Publications

Conference Full Papers and Dissertation

• Petru-Daniel Tudosiu, Walter Hugo Lopez Pinaya, Mark S. Graham, Pedro Borges, Virginia Fernandez, Dai Yang, Jeremy Appleyard, Guido Novati, Disha Mehra, Mike Vella, Parashkev Nachev, Sébastien Ourselin, M. Jorge Cardoso. Morphology-Preserving Autoregressive 3D Generative Modelling of the Brain. SASHIMI@MICCAI 2022: 66-78
• Yi Ju, Amir Raoofy, Dai Yang, Erwin Laure, Martin Schulz. Exploiting Reduced Precision for GPU-based Time Series Mining In Proceedings of the 36th IEEE International Parallel and Distributed Processing Symposium (IPDPS’ 22). 2022, Lyon, France.
• Amir Raoofy, Roman Karlstetter, Dai Yang, Carsten Trinitis, and Martin Schulz. Time Series Mining at Petascale Performance. In Proceedings of the International Supercomputing Conference 2020 (ISC’20). 2020, Frankfurt, Germany. Winnner of 2020 Hans Meuer Award.
• B Elis, D Yang, O Pearce, K Mohror, M Schulz. QMPI: A next generation MPI profiling interface for modern HPC platforms. In Parallel Computing, Volume 96, August 2020
• Dai Yang. Fault Tolerant Optimizations for HPC Systems. PhD Thesis. TUM University Press. 2020, Munich, Germany. Link
• Alvaro Frank, Dai Yang, Andre Brinkmann, Martin Schulz, and Tim Suess. Reducing False Node Failure Predictions in HPC. In IEEE 26th International Conference on High Performance Computing, Data, and Analytics (HiPC) (pp. 323-332). 2019, Hyderabad, India.
• David Jauk, Dai Yang, and Martin Schulz. Predicting faults in high performance computing systems: an in-depth survey of the state-of-the-practice. In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC’19). 2019, Denver, CO, USA.
• Bengisu Elis, Dai Yang, and Martin Schulz. QMPI: A Next Generation MPI Profiling Interface for Modern HPC Platforms. In Proceedings of the 26th European MPI Users’ Group Meeting (pp. 1-10). 2019, Zurich, Switzerland.
• Dai Yang, Tilman Kuestner, Rami Al-Rihawi, and Martin Schulz. Exploring high bandwidth memory for PET Image Reconstruction. In Parallel Computing 2019 (ParCo’19). 2019, Prague, Czech Republic. Accepted for Publication
• Dai Yang, Josef Weidendorfer, Carsten Trinitis, Tilman Kuestner, and Sibylle Ziegler. Enabling Application-Integrated Proactive Fault Tolerance. In Parallel Computing is Everwhere. IOS Press, 2018, pp. 475.

Workshop and Short Papers

• Amir Raoofy, Dai Yang, Josef Weidendorfer, Carsten Trinitis, and Martin Schulz. Enabling Malleability for Livermore Unstructured Lagrangian Explicit Shock Hydrodynamics using LAIK. In Konferenzband des PARS’19 Workshops. 2019, Berlin, Germany.
• Thomas Becker, Dai Yang, Tilman Kuestner, and Martin Schulz. Co-Scheduling in a Task-Based Programming Model. In Proceedings of the 3rd Workshop on Co-Scheduling of HPC Applications (COSH 2018). 2018, Manchester, United Kingdom.
• Josef Weidendorfer, Dai Yang, and Carsten Trinitis. LAIK: A library for fault tolerant distribution of global data for parallel applications. In Konferenzband des PARS’17 Workshops. 2017, Hagen, Germany.