Research, applied ML, and product strategy work.
Pre-trained deep learning models (Vision Transformers, ResNet) with FSL preprocessing and fusion approaches combining classical ML (SVM) with deep features for accurate Alzheimer classification.
HTML crawler that indexes university pages as documents, similarity-based query function, and ranked top-k search results.
Naive Bayes classifier with forward feature selection on the MNIST dataset, exploring feature relevance and classification accuracy.
Implementation of LDA for digit classification on MNIST, including dimensionality reduction and class separability analysis.
Face recognition pipeline on the Olivetti faces dataset using PCA for dimensionality reduction and classical learning for classification.
A cognitive science and neural data analysis project focused on processing neuronal signals, extracting spike waveforms, and applying dimensionality reduction and clustering techniques. Plotted neural information, designed a high-pass Butterworth filter, extracted spike waveforms, then applied PCA, K-Means clustering, and t-SNE for visualization.
A deep learning course project focused on Mars terrain analysis and classification using machine learning and deep learning techniques applied to planetary imagery.
An Advanced User Interface course project focused on interactive user experience, interface design, and Tangram-based interaction.
Designed a digital invoicing and smart receipt platform for SMB retailers — market analysis, SWOT, business model, retailer dashboard, mobile app, QR-based receipt collection, and loyalty tools.