I am a dual Master’s student at Purdue University (ECE & IE) with a deep interest in Machine Learning Systems, High-Performance Computing, and AI Infrastructure. I am passionate about the “full-stack” nature of modern AI—exploring the critical bridge where model architecture meets compiler design and hardware acceleration. My work is grounded in a strong architectural understanding of the deep learning landscape, from foundational CNNs and LSTMs to state-of-the-art Transformers, VLMs, and Diffusion models.
I am particularly excited by the “systems journey” of these workloads. My projects revolve around analyzing how high-level abstractions in frameworks like PyTorch are captured as computational graphs, lowered through various Intermediate Representations (IRs), and ultimately compiled down to hardware-specific instruction sets (such as PTX, SASS, or machine code). I enjoy working on challenges related to inference and training optimization, including KV-cache management, quantization, operator fusion, and Mixture-of-Experts (MoE) execution. Through my work with CUDA C++ and Triton, I actively build custom kernels that leverage tiling strategies and memory hierarchies, while also exploring scalable paradigms for distributed and disaggregated inference across diverse hardware platforms.
I am seeking opportunities in machine learning systems, infrastructure, and performance-oriented engineering roles, where I can contribute to building reliable, efficient, and scalable AI systems. This website highlights selected projects and technical writing that reflect how I approach system design, optimization, and problem-solving.