Sahana T C

An AI Approach for Stroke Types Prediction

Objective: Developed an AI- based system to predict stroke types, supporting healthcare professionals in early diagnosis and treatment planning. Project Review: Demonstrated strong predictive performance with potential for real- world integration into clinical support systems. Explored future improvements including larger datasets and advanced algorithms (Random Forest, SVM, Deep Learning). Key Features: 1. Implemented classification models (K-Nearest Neighbors, Naïve Bayes) on patient medical data. 2. Achieved high accuracy: KNN – 95%, Naïve Bayes – 96%. 3. Designed as a decision-support tool for faster and more reliable stroke identification. Technologies used: Python Scikit-learn Pandas NumPy Matplotlib /Seaborn

Mutual Fund Insight Chatbot Using LLMs

Objective: To build a system that allows users to interact with mutual fund documents through natural language queries and extract meaningful insights. Project Overview: A Python- based application that enables users to chat with mutual fund documents (e.g., reports, fact sheets). It uses LLMs and vector embeddings to retrieve relevant information and provide context-aware responses. Key Features: 1. Chat with mutual fund documents. 2. Semantic search using embeddings. 3. Context-aware AI responses. 4. Document indexing and retrieval. 5. Basic data visualization. Technologies Used: Python ( Agentic RAG, LLM integration) LangChain ChromaDB Pandas, NumPy Streamlit