A curated project portfolio across thesis research, NLP, analytics, computer vision, and applied AI.

Master thesis exploring federated learning under heterogeneous medical data distributions, with particular focus on client contribution, fairness, and the behavior of collaborative learning in realistic healthcare settings.

Bachelor thesis on automated personality recognition from asynchronous video interviews, using deep learning to infer OCEAN personality traits from visual interview signals.

A production-style internal knowledge search and Q&A system built with a Next.js frontend and FastAPI backend, using vectorless BM25 retrieval, source-grounded answers, and optional RAGAS evaluation for quality checks.

An assistive computer vision concept designed to help visually impaired users interact with their environment using recognition models.

A domain-specific named entity recognition workflow built with spaCy to extend standard NLP pipelines with custom entity labels and targeted model training.

An NLP workflow using review text to identify products through a custom spaCy model trained on domain-specific examples and iterative loss reduction.

An extractive question answering project built around BERT, focused on jointly processing a question and passage to return context-aware answers from text.

A sentiment analysis project that converts noisy Twitter text into structured signals for opinion mining and trend-level interpretation.

A collection of visual analytics work focused on transforming raw data into decision-ready dashboards and clear analytical narratives.

A large-scale exploratory analytics project examining over ten thousand Google Play apps to identify category-level patterns, growth signals, and retention opportunities.

A market exploration project analyzing cryptocurrency capitalization and ecosystem dynamics using historical Bitcoin-era market data.

A lightweight NLP utility project for improving noisy text quality through spelling correction workflows that can be reused before downstream language tasks.

A recommendation-focused analysis project aimed at connecting listeners with relevant podcast content based on behavior and preference signals.

An applied machine learning analysis for classifying human activity patterns from sensor-driven features and structured experimentation.

A movie ratings analytics and recommendation exploration project built on the MovieLens dataset to surface preference trends and personalization opportunities.

A healthcare-oriented ML project focused on early disease signal detection through feature-driven classification and model comparison.

A transformer-based text classification implementation using BERT to improve language understanding for supervised labeling tasks.

An NLP classification project for identifying spam vs. non-spam email content with end-to-end preprocessing and supervised modeling.

A practical scraping workflow in Python for collecting structured information from web sources and preparing it for downstream analytics.

A repository of NLP walkthroughs and applied examples that document practical text processing and language modeling techniques.

A vision-driven healthcare project exploring retinal image analysis for defect detection and early diagnostic support.