Projects
My work spans AI storage optimization, distributed systems, and high-throughput data processing systems. Below are some key areas/projects I have been involved with. Multi-Protocol Storage for ML Workloads Building and supporting storage systems that support both NFS (for legacy HPC workflows) and S3 (for cloud-native pipelines) over the same dataset. Key challenges include: Protocol semantic differences (POSIX vs object storage) Consistency models for concurrent access patterns Performance optimization for sequential reads (training) vs random access (checkpointing/Inference) Tier-0 Caching Strategies for GPU Clusters Researching node-local caching architectures to minimize data transfer over network fabric: ...