Clarete Crasta

Clarete Crasta

Senior Principal Engineer, HPE HPC/AI BU. Distributed Systems | HPC | Disaggregated Memory | FAM | Resiliency | Programming Models.

About me

I am a Senior Principal Engineer at Hewlett Packard Enterprise (HPE) HPC/AI Business Unit, where I lead research and development in high-performance computing, disaggregated memory architectures, and distributed storage systems. My work focuses on Fabric Attached Memory (FAM), global client-side caching for exascale HPC and AI workloads, and resilient distributed task execution frameworks – enabling applications to efficiently leverage disaggregated memory at scale.

I have contributed to 15 peer-reviewed publications in venues including CUG (Cray/HPE User Group), SNIA, Workshop on OpenSHMEM and Related Technologies, SC Workshops, and Chapel User Group (CHIUW), and hold 32 patents covering innovations in disaggregated memory, caching architectures, burst sorting, and resilient computing. I serve as Industry Track Chair for IEEE IC2E 2026 and am a SNIA Technical Council Advisor, where I review and help guide emerging storage and memory technologies.

Research Interests

  • Fabric Attached Memory (FAM) and disaggregated memory architectures
  • High-performance distributed storage systems (DAOS, OpenFAM)
  • Global client-side caching and data locality optimization for HPC and AI
  • Resiliency: fault-tolerant task execution, silent data corruption detection, work-stealing frameworks
  • Programming models for HPC and disaggregated memory (OpenFAM, OpenSHMEM, Chapel)
  • Memory-centric computing; CXL, NVMe-over-Fabrics, and persistent memory
  • Agentic AI systems: memory management for in-memory databases and query engines

News

  • [2026] Serving as Industry Track Chair for IEEE IC2E 2026.
  • [2026] Patent filed: Symmetric Heap-Based Work-Stealing Framework for Resilient Distributed Systems (Jan 2026).
  • [2025] Paper accepted at SC’25 Workshops: A Query Engine for Scientific Data Exploration using Theory, Simulation, and Artificial Intelligence Models.
  • [2025] Presented Global Distributed Client-side Cache for DAOS at CUG 2025.
  • [2025] US Patent granted: Client Allocation of Memory Across Memory Servers (Apr 2025).
  • [2025] US Patent granted: Incremental Data Backup Using a Combined Tracking Data Structure (Oct 2025).
  • [2025] Program Committee Member, HPE Tech Con 2025 & 2026.
  • [2024] US Patent granted: Client Update of Data Modification Tracking Structure (Oct 2024).
  • [2024] Presented FAM Support in OpenSHMEM at Cray Users Group Australia 2024.