An Nguyễn Thanh

Designed, deployed, and optimized an intelligent RAG-based AI agent delivering high-performance document retrieval and analysis. - Engineered a RAG pipeline in Python, leveraging large language models to boost document search accuracy and response relevance. - Implemented a scalable FAISS vector database, enabling semantic search across documents while maintaining low memory overhead. - Containerized and deployed the solution on AWS, achieving seamless scalability for enterprise-level data workloads.

Developed and deployed high-performance RAG pipeline through multimodal retrieval and optimized large language model inference. - Built a multimodal RAG pipeline with LangChain framework, enabling fast and accurate information extraction from large-scale document partitions. - Optimized LLM inference using vLLM and MLOps practices, reducing GPU memory usage while maintaining model throughput and enhancing scalability. - Developed and integrated an MCP server with an autonomous web-crawling AI agent capable of real-time data enrichment from online sources, ensuring continuous knowledge updates.

Developed and optimized large language model–powered systems for intelligent data protection and performance benchmarking. - Fine-tuned GPT-2 using PyTorch and TensorFlow to build an AI-driven data protection system, improving threat-detection accuracy. - Designed and deployed a secure MySQL data infrastructure to manage cybersecurity-sensitive datasets. - Benchmarked multiple LLMs from Hugging Face, analyzing latency, perplexity, and throughput to identify the optimal trade-offs for production deployment.