Kenneth Ng.
I enjoy building intelligent digital solutions.

VectorShop is a sophisticated semantic search system for small-to-medium sized online stores, allowing customers to find products using natural language queries instead of just keywords. The system combines keyword search, vector similarity, and AI reasoning to deliver highly relevant search results even for complex queries.

The system uses a hybrid architecture that combines sparse retrieval (BM25) with dense vector embeddings (FAISS) and a reranking layer using transformer models. This approach delivers sub-100ms query times even with millions of products, making it suitable for production environments.

This project uses machine learning to predict SpaceX Falcon 9 first stage landing success, which helps estimate launch costs since SpaceX rocket launches are significantly cheaper when the first stage can be reused. The analysis combines data from multiple sources, including the SpaceX REST API and web scraping from Wikipedia.

The project achieved 83% prediction accuracy by analyzing launch records with details on boosters, payloads, and landing outcomes. It includes interactive visualizations and dashboards to explore the data and understand the factors that influence landing success.

The Generative AI Blockchain Security Analysis System leverages artificial intelligence to analyze blockchain transactions and smart contracts for security vulnerabilities and potential fraud. This project was developed for the Chainlink Hackathon 2024 and implements a chain-agnostic approach supporting Ethereum, Polygon, and Avalanche networks.

The system combines Chainlink Functions for secure on-chain data retrieval with OpenAI's LLM for security analysis, providing detailed reports on potential vulnerabilities and suspicious transactions. Analysis reports are stored on IPFS via QuickNode for decentralized, permanent storage.

The AI Education System is a platform that leverages generative AI to create personalized blockchain and Web3 learning experiences. Built for the Theta Network 2024 Hackathon, this system demonstrates how AI can revolutionize education by adapting to each learner's background, interests, and learning style.

The platform uses large language models (LLMs) to generate customized course content, answer student questions, create quizzes, and provide explanations—all tailored to the individual learner's profile. It features blockchain-based authentication using Theta wallet addresses and integrates with Theta EdgeCloud for AI model inference.