Paper Title: Performance at Scale for Model Training

Paper Abstract: Model training for complex deep neural networks (DNNs) is becoming a major issue as use cases transition from computer vision to multi-modal and conversational AI. Storage I/O creates a huge bottleneck as the compute layer becomes both more complex and more parallel. A low-latency, high-throughput parallel file system is an essential part of the solution. It both keeps computation times reasonable and offers scalability to handle ever-larger datasets.

Paper Author: Shailesh Manjrekar, Head of AI, WekaIO

Author Bio: Shailesh Manjrekar is responsible for WekaIO's AI strategy and technical alliances, aligning Weka's product roadmap and validating the software on key platforms in our target markets and use cases, such as Artificial Intelligence (AI) and Machine Learning (ML), genomics, finance, and high-performance computing (HPC). Before joining Weka, Shailesh was head of AI at SwiftStack, where he was responsible for product, solutions, and corporate development, before NVIDIA acquisition. Prior to SwiftStack, he held roles at Vexata, EMC, NetApp, Brocade, Force10 Networks, and he also held positions at Hewlett Packard and Aarohi Communications. Manjrekar holds an MBA from San Jose State, College of Business, Certification in Mastering Project Management from Haas School of Business, UC Berkley, and a B.S. in Electrical Engineering from the University of Mumbai.