My Projects

Here's a collection of my recent work

Churn Predictor

Compact end-to-end churn prediction pipeline for telecom. Identifies high-risk customers using robust preprocessing, feature engineering, XGBoost (with hyperparameter tuning) and SHAP explainability to guide retention actions.

HealthRAG

Retrieval-Augmented Generation (RAG) chatbot for medical document QA. Uses PDF ingestion, FAISS vector store, LangChain and a hosted LLM (Hugging Face) to answer queries based on provided medical documents. Includes Docker support and a Flask demo.

Smart Vehicle Detector

AI-powered real-time vehicle detection pipeline combining YOLO-based object detection with a Vision-Language Model (VLM) for fine-grained vehicle classification and structured JSON outputs for downstream systems

CineAPP (SDK)

Python SDK for CineAPP REST API — lightweight client that returns Pydantic models, dict or Pandas DataFrames, designed for data analysis workflows and easy integration with the CineAPP API

UNETR 2D - Medical Segmentation

From-scratch PyTorch implementation of a 2D variant of the UNETR architecture for medical image segmentation. Focus on reproducible training, experiments and tooling for biomedical datasets

EasyNodule

Desktop application for AI-driven pulmonary nodule classification from segmented CT scans — combines our custom CapsNet model with patient management, secure accounts, PDF report generation and a statistics dashboard