Shafin Ahmed
Machine Learning | Bioninformatics | Research | Teaching
BRAC University
Project
Deep generative framework to integrate RNA, ATAC, proteomics and metabolomics data where each dataset has only some modalities. Uses modality-specific encoders and decoders with a shared latent space, plus missing-not-at-random (MNAR) modeling. Adds calib
Project
Built a pipeline to classify breast cancer samples from high-dimensional RNA-seq data (TCGA-BRCA). Used log transform and mutual-information feature selection, then compared Logistic Regression, XGBoost and a Neural Network under class imbalance.
Project
Built an end-to-end workflow covering EDA, cleaning, encoding, scaling and supervised modeling. Handled class imbalance and compared models on accuracy, precision, recall and ROC-AUC, with a KMeans baseline.
Project
Evaluated Platt scaling and isotonic regression for tree-based models on binary and multiclass healthcare tasks. Used the CDC Diabetes Health Indicators 2023 data with ROC-AUC, Brier score, ECE and reliability diagrams.