I am currently a Deep Learning Engineer on the cuDNN team at NVIDIA working on accelerating deep learning workloads on GPUs. I'm broadly interested in performance engineering, machine learning systems, hardware architecture, and empirical ML. Currently also writing a MLSys Review newsletter

I graduated from Massachusetts Institute of Technology (MIT) with a B.S. and M.Eng (coursework here). where I worked on understanding the performance of large language models for my thesis.

Some things I worked on in the past:

• NVIDIA: added CUDA kernels for Graph Neural Networks (GNNs) into the cuDNN library
• MIT Han Lab: developed high-performance computer vision kernels for TinyML on edge devices
• MIT-IBM Watson AI Lab: accelerated distributed training on HPC clusters for large scale graph neural networks [MLSys 2022 paper]
• Prod Cohort 2.0: attempted doing startups and an AI newsletter
• Cadence Design Systems: developed a machine learning workflow for faster and more accurate VLSI circuit timing analysis
• Praxis: benchmarked diffusion models and reviewed the torch.distributed package
• Matrix Capital Management: venture scout for crypto/web3 companies and projects
• Trail of Bits: performed formal verification studies on the Ethereum 2.0 CBC Casper protocols
• MIT Media Lab Digital Currency Initiative: implemented simulations of consensus algorithms to study 51% attack mitigation and symmetry breaking
• MIT CSAIL Decentralized Information Group: analyzed statistical properties of transactions on the bitcoin blockchain empirically
• MIT Bitcoin Club: vice president; content and marketing director of the MIT Bitcoin Expo
• Alphastar Academy: taught USA Computing Olympiad preparation courses

Outside of CS, I enjoy history of art and architecture, photography, and classical music.

Resume available upon request. If you have any questions or would like to chat, feel free to reach out (social links below).

(PGP public key fingerprint: 07DF 3BA8 4259 DF0A 73E0 365B 5A80 97CE 8F78 0EB7)