Research Projects

Explore our cutting-edge research in machine learning for scientific discovery, spanning biology, economics and physics.

Lattice: Structured Multi-Agent Code Generation
Engineering
Automated Software Engineering with Coordinated AI Agents

Brandon Yee, Benjamin Pellegrini, Kundana Kommini, Arnva Sharma

Submitted to AAAIWSubmitted to IEEE TSE

A structured multi-agent framework for automated code generation and software engineering tasks, demonstrating coordinated AI systems for complex software modifications.

Machine Learning-Driven Prediction of TLR4 Binding Affinity
Biology
A Comprehensive Molecular Feature Analysis for Drug Discovery

Brandon Yee, Max Rutkowski, Wilson Collins

Submitted to CSB Review

Advanced machine learning models for predicting TLR4 binding affinity using comprehensive molecular feature analysis to accelerate drug discovery processes.

TLR4 Binding Data
Biology
Comprehensive dataset for TLR4 binding research

Max Rutkowski, Brandon Yee, Daniel Huang, Lev Kung, Oliver Pierborne

Submitted to JMGM

Curated dataset containing TLR4 binding affinity data, available on both GitHub and Kaggle for reproducible research.

When Does Quantitative Easing Work?
Economics
Threshold Effects, Market Distortions, and International Transmission

Brandon Yee, Jonah Rothlein

Working PaperSSRN: 5445554

Comprehensive analysis of quantitative easing policies, examining threshold effects and international transmission mechanisms using advanced econometric methods.

Prometheus: Unsupervised Discovery of Phase Transitions
Physics
Order Parameters in the 2D Ising Model Using Variational Autoencoders

Wilson Collins, Brandon Yee, Caden Wang, Mihir Tekal

Submitted to AAAI-26 SAPPSubmitted to AAAIW-26 AI2ASESubmitted to Physics Review E

Novel application of variational autoencoders for unsupervised discovery of phase transitions and order parameters in statistical physics models.

Meta-Learning Physics-Informed Neural Networks
Physics
Few-Shot Parameter Inference

Brandon Yee, Wilson Collins, Benjamin Pellegrini, Caden Wang

Submitted to AAAI-26 SAPPSubmitted to AAAIW-26 AI2ASESubmitted to CMAMEPresented at Brown Crunch Labs

Advanced meta-learning framework for physics-informed neural networks enabling efficient few-shot parameter inference in physical systems.