Research, engineering, and building.
Researcher • Present
Investigating emergent behaviors in large language models and training paradigms to accelerate reasoning. Designing experiments on curriculum learning and phase transitions in LLM development. Also exploring the cross-modality of symbolic reasoning within large vision-language models. Presented poster at Technigala Symposium on a novel robustness paradigm for worst-class accuracy in vision models.
Research Intern • Present
Interpreting graph neural networks and multi-agent debate frameworks. Co-authoring a paper on question-answer and single-agent winning prediction using node embeddings and signed convolutional graph networks. Currently beating benchmarks by 7% on question-answer and >20% on single-agent winning prediction.
Machine Learning Engineer • Present
Deployed full-stack ML applications with real-time inference for a myriad of contractors. Built a barnacle counting system for the National Park Service using SOTA object detection models. Built a RAG system for enhanced context extraction at a YC-backed code review startup (Benchify). Deployed a CNN for plastic classification for the Dartmouth Chemistry Department, which worked with excruciatingly poor data quality. For fun, I run the lab's hiring process - managing 30+ interviews and 150+ applications.
Engineering Intern • 2024
Prototyped human-in-the-loop ML paradigm for spoken language verification with 200x better accuracy than traditional training, using orders of magnitude less data in training. Collected a large corpus of data on 12 South Asian languages. Configured 8 Intel NUC servers and optimized 2 ML algorithms for CPU performance.
Engineering Intern • 2022
Constructed AI surrogate models of scientific code for the Parallel Hardware viA Surrogate Models project. Architected physics-informed neural networks achieving MSE <.001 for approximating differential equation solutions. Wrote bash scripts to reduce and automate installation steps across various operating systems. Rapidly self-taught differential equations and PyTorch for deployment.
Astericks denote graduate level courses.
Object-Oriented Programming and Data Structures • Discrete Mathematics • Algorithms • Software Design and Implementation • Full Stack Web Development • Security and Privacy* • Digital Electronics • Machine Learning* • Deep Learning* • Deep Learning Seminar on Robustness and Generalization* • Vector Calculus • Honors Linear Algebra • Number Theory • Discrete and Probabilistic Systems • Intro to Engineering
Ordinary Language Philosophy • Philosophy and Computers • Introductory Economics • Public Policy • Writing Seminar (The Art of Human Dialogue) • Expository Writing (Ownership of Ideas)
Awards, honors, grants, writings, etc.
Lovelace Scholar • Neukom Scholar • Kemeny Prize 2025 • Presidential Scholar • Founder's Grant • MIT Sandbox Grant • Dunkin Donuts DMV Scholar • Riverbend High School Valedictorian • National Rural Scholar • Newspaper Writings