Deep Learning Portfolio

A collection of deep learning projects spanning computer vision, time series forecasting, medical imaging, mobile AI, and agentic RAG systems.

Projects

Project	Domain	Technique	Stack
Image Classification	Computer Vision	CNN (Conv2D + MaxPooling)	TensorFlow/Keras, Fashion MNIST
Stock Price Prediction	Finance	LSTM (Recurrent Neural Network)	TensorFlow/Keras, yfinance, Streamlit
Pneumonia X-Ray Classifier	Medical Imaging	CNN (Binary Classification)	TensorFlow/Keras, Streamlit
Smart Mobile App	Cross-Platform AI	Firebase Auth + TFLite Inference	Flutter, Dart, Firebase
Agentic AI Backend	Conversational AI	RAG + MCP Protocol + Tool-Calling	FastAPI, Ollama (Llama 3.2), Python

Project Details

Lab 1 — Image Classification (CNN)

Trains a convolutional neural network on the Fashion MNIST dataset (60K images, 10 classes). Exports models in both Keras H5 and TensorFlow Lite formats for server and mobile inference.

Lab 2 — Stock Price Prediction (LSTM)

Fetches live stock data from Yahoo Finance, builds sliding-window sequences, and trains a two-layer LSTM network. Interactive Streamlit UI for visualization and forecasting.

Lab 3 — Pneumonia X-Ray Classifier

Streamlit web app for uploading chest X-ray images and classifying them as Normal or Pneumonia using a trained CNN model. Includes confidence scoring.

Lab 4 — Smart Mobile App (Flutter)

Cross-platform Flutter app with Firebase authentication, an AI chatbot connected to the MCP backend, and on-device image classification using TensorFlow Lite.

Agentic AI Backend (MCP Server)

FastAPI server implementing the Model Context Protocol (MCP) with RAG-based knowledge retrieval, tool-calling capabilities, and Ollama LLM integration.

Tech Stack

Category	Technologies
ML/DL	TensorFlow, Keras, NumPy, Pandas, scikit-learn
Web	Streamlit, FastAPI
Mobile	Flutter, Dart, Firebase
AI Backend	Ollama (Llama 3.2), MCP Protocol, RAG
DevOps	Docker, Docker Compose

Quick Start

# Lab 1: Train CNN
cd Lab1_Image_Classification && pip install -r requirements.txt && python train_model.py

# Lab 2: Stock Prediction UI
cd Lab2_Stock_Prediction && pip install -r requirements.txt && streamlit run app.py

# Lab 3: Medical Image Classifier
cd Lab3_Model_Deployment && pip install -r requirements.txt && streamlit run app.py

# Lab 4: Flutter App
cd Lab4_Smart_App_Flutter/smart_app && flutter pub get && flutter run

# Agentic AI Backend
docker-compose up -d  # Start Ollama
cd Control_Backend_AgenticAI && pip install -r requirements.txt && python main.py

License

MIT License. See LICENSE for details.