Projects

Selected things I have built from LLM-powered apps to distributed data systems.

Projects

LLM · RAG · Mobile

Thea: LLM-powered theater ticketing

Cross-platform ticketing app (Flutter + Flask) with an integrated LLM assistant (Llama 3.2 3B) and a RAG pipeline for real-time schedules, seating charts, and recommendations.

RAGLangChainFlutterFlask
GenAI Hackathon · SDGs

Skin lesion screening app

Prototype mobile app screening skin lesions from smartphone photos: a VAE for feature extraction and anomaly scoring (PyTorch, OpenCV), with explanations generated by Gemini 1.5 Pro.

VAEPyTorchOpenCVXAI
Novartis Datathon 2025

Generic erosion forecasting

Developed machine learning models to forecast pharmaceutical sales volume after generic market entry using longitudinal country–brand data. Performed exploratory data analysis, feature engineering, and time-series forecasting to model generic erosion under two business scenarios in the 8th Novartis Datathon.

Machine LearningTime SeriesForecastingPython
LLM · Data Analytics

Insight Bot

AI-powered analytics assistant that enables natural language interaction with structured datasets. Built with LangFlow, GPT, and Streamlit, providing conversational data exploration, real-time insights, and persistent chat history through an intuitive web interface.

LLMsLangFlowStreamlitERP Systems
SemEval 2025

Food hazard detection

NLP pipelines for food hazard detection using entity recognition and transformer models (BERT, RoBERTa) for the SemEval 2025 shared task.

TransformersHugging FaceNER
Distributed Systems

Big data management systems

Containerized full-stack applications on Redis, MongoDB, Neo4j, and Azure Analytics for real-time systems, analytics, and distributed computing.

RedisNeo4jMongoDBDocker