Ph.D. Candidate, Computer Engineering, University of Toronto
I research efficient AI systems, with a focus on compression, quantization, data movement, and hardware/software co-design for bandwidth- and memory-constrained deployment.
I am a Ph.D. candidate in Computer Engineering at the University of Toronto, advised by Prof. Andreas Moshovos. My work sits at the intersection of machine learning systems, computer architecture, and practical AI deployment. Recent projects span tensor compression, low-precision representations, training-efficiency evaluation, computational graphics systems, and responsible human-in-the-loop AI workflows.
Compression and quantization-aware methods for ML workloads where data movement is a bottleneck, with emphasis on practical representations that reduce bandwidth and memory pressure without changing model behavior.
Representative: Shannonic; Atalanta; Every Bit Matters
Systems work that connects compression algorithms and quantized data formats to real implementation constraints, including codec state, cache behavior, throughput, hardware cost, and deployment across memory, network, and edge–cloud links.
Representative: Shannonic; Atalanta
Evaluation methods for efficient training systems, with emphasis on fair comparisons, reproducible baselines, and metrics that reflect practical end-to-end training behavior.
Representative: Selective Workload Skipping (IHIET‑AI 2025)
Systems work on efficient graphics pipelines, with emphasis on compression, latency, quality, and resource-aware deployment for emerging visual representations.
Representative: COGS; Gaussian splatting systems
Human-in-the-loop AI-assisted grading workflows for large programming courses, focusing on rubric alignment, grading consistency, human oversight, and responsible deployment.
Current work: AI-assisted grading for large programming courses
Broader systems work on on-device neural interfaces, with emphasis on reducing resource requirements while preserving application-level quality.
Representative: Marple (ASPLOS 2024)