AI Engineer with LLM, RAG, and Agentic Systems

Ha Ngu Long Nguyen

AI Engineer

Thu Duc, Ho Chi Minh City 094******* linkedin.com/ longnguyenha555@in/nguyen-ha-ngu-gmail.long com github.com/longnguyenha050 Summary

AI Engineer focused on building LLM, RAG, and Agentic systems. Delivering value by transforming complex AI re- search into cost-effective, production-ready solutions that prioritize reliability and performance. My goal is to architect autonomous AI Agents that drive high-precision, real-world impact for enterprise applications. Education

University of Information Technology – UIT 2022 - 2026 Major in Computer Science

GPA: 3.2/4.0 UIT Global Scholarship recipient

Technical Skills

Core AI Concepts: Deep learning, Machine Learning, Computer Vision, RAG, Information Retrieval, LLM Programming Languages: Python, C++, JavaScript, HTML/CSS Databases: MongoDB, Cloudinary, ChromaDB (Vector DB), PostgreSQL Web & Development Tools: Node.js, React.js, GitHub, Jupyter Notebook, FastAPI. Cloud Platforms: Docker, Linux, Familiar with Google Cloud Platform (Vertex AI, Cloud Run, Cloud Storage). Experience

AI Engineer Intern - TMA Solutions 9/2025 - 12/2025 Sign Language Recognition Project

• Participated in a skeleton-based Sign Language Recognition project using sequential pose data.

• Optimized Next-Word Prediction by fine-tuning pre-trained Transformer architectures (ViT5 & PhoBERT) with under 1B parameters to improve real-world execution accuracy. Projects

Video Event Retrieval & Tracking System Backend — Frontend

• Developed a robust semantic search pipeline utilizing CLIP embeddings and FAISS indexing to enable efficient natural language queries.

• Integrated multimodal AI architectures for scene boundary detection, YOLOv8 for object detection.

• Implemented OCR solutions using VietOCR and PaddleOCR to extract and index textual metadata from video frames.

Advanced RAG Chatbot for Shoe E-commerce Platform GitHub

• Architected an advanced RAG system using LangChain and LangGraph to power an AI shopping assistant for a shoe e-commerce platform.

• Developed a multi-retriever architecture integrating ChromaDB (Vector Retriever - Hybrid Search), Tavily API

(Internet Retrieval), and MongoDB (MongoDB Retriever) to optimize context retrieval.

• Applied Cross-encoder Reranking to refine retrieved results from both vector and internet sources, enhancing context precision before LLM generation.

AI-Powered Adaptive Mock Interview Agent Backend — Frontend

• Engineered a Multi-Agent Interviewing system, enabling targeted evaluation across Project, Behavior, and Skill domains, improving assessment precision by 40%.

• Architected a hybrid memory (STM/LTM) via Redis, reducing state-retrieval latency for seamless real-time conversa- tion tracking.

• Optimized LLM resource allocation via Redis caching, successfully cutting API operational costs while maintaining 24/7 responsiveness.

LLM Hallucination Detection System GitHub

• Fine-tuned PhoBERT-base for a 3-class classification task (No, Intrinsic, Extrinsic hallucinations).

• Developed a sentence-level retrieval system that decomposes input context into individual sentences, utilizing Embedding- based similarity to cross-reference each segment against the query.

• Attained a significant classification Accuracy of 0.823 on a complex Vietnamese hallucination dataset. Certificates

Google Data Analytics (Coursera) AI for Anomaly Detection (NVIDIA) IELTS 6.5 (British Council)

Contact this candidate