Revolutionizing medical process.
Selected Projects
This repository features an XGBoost baseline, along with 5 independant GNN-variant models, aiming to predict the pKd value between a given drug and it's target protein.
Model that predicts diabetes onset based on several health parameters typically found in medical records.
Scrapers to autocheck for company internship postings.
Chatbot tailored towards mental health support using retreival augmented generation, OpenAI API.
Uses Meta's ESM-2 protein language model to tweak input protein and return a slightly modified one with a high likelihood of similar properties.
Work Experience
Blog
Predicting Drug-Protein Binding Affinity with Graph Neural Networks
Building an end-to-end machine learning pipeline using Graph Neural Networks to predict how strongly drug molecules bind to proteins, with lessons on data splitting, protein embeddings, and molecular graph representations.
HydraLA-Net: Multi-Lesion Segmentation for Diabetic Retinopathy Detection
Exploring preprocessing and DL methods for multi-lesion segmentation in diabetic retinopathy detection, including a novel HydraLA-Net architecture and insights on handling class imbalance and data augmentation.
Progressive Optimization of HydraLA-Net for Microaneurysm Segmentation
Research paper submitted to CUCAI 2026 detailing my Wat.AI team's project, See-DR. The paper details our entire process, from motivations and methods, to results, and interpretations.
Connect
Feel free to contact me at s78shah@uwaterloo.ca