I'm a CS undergrad at Rutgers studying applied ML and robotics. I do research in two labs: PRACSYS, where I work on robot controller verification using flow matching (first-author paper under review at IROS), and TRUST-ME, where I work on genomic sequence classification with a novel attention mechanism on DNABERT-2 (second-author paper under submission). I'm an Accenture ML Fellow through Cornell Tech and I lead the software department for Rutgers VEX U Robotics.
I got into building through FIRST Robotics in high school, then VEX U in college. Somewhere along the way I stopped thinking about robots as machines and started thinking about what it takes for any AI system to work reliably in the real world. That pulled me into research labs, generative modeling, multi-agent systems, and eventually a habit of building things faster than I can write about them.
I gravitate toward problems where the system has to actually work, not just demo well. That's why I flew to SF to build a latte art robot with a VLA, built a voice agent that handles real restaurant calls end to end, and wrote a shared memory library for multi-agent systems that cuts redundant work in half. I don't pick projects by domain. I pick them by whether the problem is hard enough to be interesting and real enough to matter.
Robotics ML. Regime classification in high-dimensional dynamical systems under Prof. Kostas Bekris. ~60% fewer labels needed, 97% classification accuracy near regime transition points.
Computational genomics under Prof. Yossi Cohen. Transformer sequence models on 640M+ DNA bases for fungicide resistance prediction. ~73% held-out accuracy.
NLP for misinformation detection. Transformer pipelines across 170K+ news articles. ~20% accuracy lift, ~35% shorter test cycles. Placed via Cornell Tech Break Through Tech AI.
Applied ML fellowship with industry placements. Selected from 3000+ applicants.
22nd internationally at VEX U Worlds 2025.