AI Projects by Htet

Final year project Brunel University

📌 XAI 🎓 Final Year Project This project was developed as part of my undergraduate third year final projectdegree at Brunel University London. 📖 Project Description Multi-AI Electrocardiogram Classification Framework Develop, assess, and implement a comprehensive, multi-model AI system for ECG categorization and ECG-related inquiries. Comparative Examination of CNN Parameters* Systematically examine how CNN-based ECG classification performance is affected by dropout, learning rate, batch size, and layer count. Data and Signal Processing Insight* Assess the impact of oversampling, signal sampling rate (100 Hz vs 500 Hz), denoising techniques, and feature representation methods (raw ECG, FFT, CWT, RR intervals, PQRST morphology) on model correctness, F1-score, loss, and stability. Model Development and Evaluation Design and evaluate the performance of many ECG classification models, including 1D CNN, ResNet1D, LSTM, GRU, Transformer, Capsule Network, and CWT-based explainable ResNet50. Evaluate prediction accuracy, robustness, interpretability, computational efficiency, and ecological effect. Reliability Verification* To determine the consistency, dependability, and credibility of the best-performing models, do ten-run repeated validation. ECG-Specific Language Model Creation* Construct an ECG-focused big language model pipeline that consists of: Lora-based fine-tuning of LLaMA-2. Comparative analysis with quantized LLaMA-3. Assessment of retrieval-augmented generation (RAG) for medically precise question answering. 7.Automated Generation of Question-and-Answer Datasets* Create an agent-based RAG system using Lang Chain/Lang Graph for the automatic generation of Q&A datasets from online sources, hence reducing human labour. Production-Ready Full-Stack Platform Integrate and implement a scalable ECG question-and-answer and assessment platform using Next.js, Fast API, Redis, PostgreSQL, Docker, and AWS. 🛠️ Technologies Used Python / JavaScript / HTML/ CSS. Lang Frameworks / Libraries (e.g., FastAPI, React, TensorFlow) Chain/ Lang graph/ chrome graph/ next.js/ fast api/ AWS /Docker 👤 Author HTET PAING AUNG 🎓 Supervisor Professor Tatiana Kalganova 📄 License This project is licensed under the MIT License.

How to use it: 1. Enter a URL 2. Click Evaluate 3. Click Generate 4. Download the generated Q&A pairs. Processing time depends on the size and complexity of the content.

Customer Support Q&A Chatbot

A LLaMA 2 model fine-tuned on a custom customer support Q&A dataset for chatbot applications in business and e-commerce environments.

AI for Lung X-ray Analysis

A CNN-based medical imaging project that detects COVID-19, pneumonia, tuberculosis, and normal conditions from lung X-ray images.

ECG Signal Classification with AI

An ECG classification project using LSTM models trained on the PTB-XL dataset. The model performance varies across classes, with F1 scores ranging from approximately 60% to 90%.

YOLO Object Detection Transfer Learning

A transfer learning project using YOLO for real-time object detection tasks. It can be adapted for use cases such as robotics, automation, and defense-related detection systems.

Handwritten Digit Classifier

A handwritten digit recognition project based on the MNIST dataset using edge detection and convolutional neural networks.

Tic Tac Toe Game AI

A game AI project exploring CNN, LSTM, and hybrid models for strategy learning in Tic Tac Toe. This project compares model behavior and performance across multiple architectures.

Rock Paper Scissors with LSTM

A fun AI game project that uses LSTM-based move prediction for Rock Paper Scissors gameplay.

Smart AI Web + Research Assistant

This research assistant combines multiple tools to provide accurate, context-aware responses in a RAG-style workflow.

• Wikipedia for general knowledge queries
• Semantic Scholar API for academic and research questions
• DuckDuckGo for real-time web search
• Embedding Memory with FAISS and Sentence Transformers
• LLaMA 3.1 via Hugging Face for response generation

DeepSeek AI Agents and Automation Projects

A collection of 25 AI agent and automation project ideas built around Python, automation workflows, and LLM-based tooling.

• Project 1: Text Summarizer AI
• Project 2: Text Generator AI
• Project 3: Grammar and Spelling Checker AI
• Project 4: Named Entity Recognition AI
• Project 5: Sentiment Analysis AI
• Project 6: Customer Support AI
• Project 7: Personal AI Assistant
• Project 8: Legal AI Assistant
• Project 9: Medical Symptom Checker AI
• Project 10: Product Recommendation AI
• Project 11: Auto Email Responder
• Project 12: Resume Generator AI
• Project 13: Meeting Minutes Generator AI
• Project 14: Auto PDF Extractor
• Project 15: Content Writer AI
• Project 16: Code Compiler AI
• Project 17: SQL Query Generator
• Project 18: Code Debugger AI
• Project 19: AI-based Document Generator
• Project 20: AI-powered API Tester
• Project 21: Customer Feedback Analyzer AI
• Project 22: Real-Time News Summarizer AI
• Project 23: Financial Analyzer AI
• Project 24: AI-based Job Application Screener
• Project 25: AI Research Paper Summarizer