About
I'm a Computer Science & Mathematics student at MVSU to graduate in May 2026.
Experience at ActiveCampaign and Science Gateway.
Selected Projects
LSM-DB: C++ Log-Structured Merge-Tree Storage Engine | Project Link
- Developed a high-throughput NoSQL engine (similar to RocksDB) using LSM-Tree architecture for durable data persistence.
- Implemented a SkipList-backed MemTable with a Write-Ahead Log (WAL) to ensure zero data loss during crash recovery.
- Optimized read paths using Bloom Filters for negative lookups and a Sparse Index for accelerated disk access.
- Built a background Compaction Engine using K-way merge sort to consolidate SSTables and manage the data lifecycle efficiently.
FlashAttention-Style Transformer Kernel | Project Link
- Added CUDA C++ implementation of FlashAttention, optimizing memory usage and performance.
- Introduced Online Softmax algorithm to reduce memory complexity from O(N²) to O(N).
- Developed forward and backward kernels with support for half-precision (FP16) inputs.
- Implemented CPU reference for FlashAttention for validation against PyTorch's SDPA.
- Created Python interface for autograd compatibility and ease of use in PyTorch.
- Added profiling scripts to benchmark performance and efficiency of the CUDA implementation.