MIDS Capstone Projects: Summer 2026

Our hybrid HEA and HEC ML model leverages applicable physics knowledge transfer between the two different material classes. With this advantage, experimental…
Integrating open-source Hydrologic Engineering data and Physics-Informed Neural Networks to predict structural % damage from localized flooding.
Echoes of the Heart shows that synthetic latent datasets significantly improve the performance of diagnostic models for echocardiography.
AI-powered guitar transcription tool that converts audio recordings into accurate, playable tablature with standard notation, timing, and key.
MacroIQ: agentic recipe recommendations grounded in USDA data, FoodOn tags, and dietary rules, not LLM guesses.
Nutritionell - Fast and Transparent Nutrition Intelligence, One Grocery Photo at a Time
OmniSearch: A Simulation Platform for Coordinated Drone and Ground Robot Swarms Deployments in Wildfire Search and Rescue
Can you trust an AI mind-reader? A Berkeley MIDS capstone auditing whether AI interpretability tools read a model's mind — or just guess.