MY STORY
A short, honest narrative — the kind recruiters actually read. Research + systems + taste for clean design.
Timeline
I build reproducible KD pipelines, measure GPU/CPU power, and connect performance retention to energy-per-token. I like research that ships.
A Single Board Cluster Competition challenges student teams to design, tune, and operate a parallel SBC-based HPC system by running standard benchmarks like HPL and HPCG, alongside optimizing a mystery parallel simulation application that tests real-world performance, scalability, and system engineering skills. Placed 2nd
The Winter Classic Invitational Student Cluster Competition is a virtual, mentor-guided HPC competition where top student teams optimize real scientific applications on national-lab-scale systems and showcase both performance and presentation skills. Team Matador: Placed 3rd
Web tooling, internal apps, and learning to design interfaces that feel intentional — not accidental.
What I optimize for
Current focus - Knowledge Distillation Methods on LLMs
Publishing strong results around energy-efficient distillation that proves I can do both: research rigor and production-quality engineering.
Side Project - Reliable Cluster Managment
Publishing robust results in reliable cluster management while designing, esting, and deploying a custom monitoring system—demonstrating both research rigor and production-grade engineering through end-to-end system design, hardware-level coding with 20+ I²C devices, and full system deployment.
I'm most interested in work where performance claims come with real system evidence.
I also play a role in reliable cluster management where operational claims are backed by concrete, end-to-end system evidence.
