Skip to content
View HuNguyen51's full-sized avatar
  • Ton Duc Thang University
  • Ho Chi Minh city, Viet Nam
  • 11:09 (UTC +07:00)
  • LinkedIn in/hunguyen0501

Block or report HuNguyen51

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
HuNguyen51/README.md

hunguyen51


Xin chào 👋, mình là Nguyễn Hưng

🤖 AI Engineer · Xây dựng hệ thống thông minh · Việt Nam 🇻🇳


  • 🏗️ Đang xây dựng ScanX — nền tảng AI doanh nghiệp theo kiến trúc microservices, cung cấp các AI engine có thể tái sử dụng cho nhiều ứng dụng thực tế.

  • 🧠 Mình làm gì: Thiết kế và triển khai các hệ thống AI production-ready trong các lĩnh vực Computer Vision, NLP, Speech RecognitionMulti-Agent Systems — từ nghiên cứu đến vận hành thực tế.

  • 🔬 Những dự án nổi bật:

    • Phát triển chatbot hỗ trợ trải nghiệm khách hàng với kiến trúc Multi-tenant, kết hợp RAG (Qdrant, VoyageAI) và LangGraph để điều phối các Agent có trạng thái, hỗ trợ upsell và tra cứu menu thông minh.
    • Xây dựng async pipeline phân tích cuộc họp (Speech-to-Text + Speaker Diarization + LLM extraction) với RabbitMQ, Redis & LangChain
    • Thiết kế kiến trúc hệ thống camera an ninh thông minh nhận diện khuôn mặt & phát hiện đối tượng theo thời gian thực (YOLO11 + DeepSort + MTCNN)
    • Nghiên cứu Multi-Agent Systems với Google A2A & MCP — các agent chuyên biệt được điều phối bởi Host/Manager Agent
    • Ứng dụng GCN (Graph Convolution Networks) vào bài toán khám phá thuốc, dự đoán hoạt tính sinh học
    • Xây dựng mô hình phát hiện ngôn ngữ thù địch tiếng Việt với PhoBERT & LSTM
    • Triển khai phân cụm K-Means và tối ưu hóa hơn 10 thuật toán Machine Learning (SVM, Random Forest, AdaBoost) cùng Deep Learning (Keras), đạt độ chính xác 94.74% trong việc dự báo.
  • Điểm khác biệt: Mình không chỉ train model — mình thiết kế toàn bộ hệ thống: async pipeline, containerized services, scalable API và agent orchestration. Mình đặt trọng tâm vào hiệu năng, khả năng tái sử dụng và giá trị thực tế.

  • 🌱 Đang khám phá: Context Engineering · Advanced RAG · Multi-Agent Orchestration · MCP Protocol

  • 👨‍💻 Tất cả dự án của mình: github.com/HuNguyen51

  • 💬 Hỏi mình về: LLMs · Computer Vision · Multi-Agent Systems · AI System Design

  • 📫 Liên hệ: hunguyen3525@gmail.com

Hi 👋, I'm Nguyen Hung

🤖 AI Engineer · Builder of Intelligent Systems · From Viet Nam 🇻🇳


  • 🏗️ Currently building ScanX — an enterprise AI platform with microservices architecture, providing reusable AI engines for real business applications.

  • 🧠 What I do: I architect and ship production-ready AI systems across Computer Vision, NLP, Speech Recognition, and Multi-Agent Systems — from research to deployment.

  • 🔬 Recent highlights:

    • Developed a F&B customer assistant using a multi-tenant architecture, featuring RAG (Qdrant + VoyageAI) and stateful LangGraph agents for menu discovery and automated upselling.
    • Built an async meeting analysis pipeline (Speech-to-Text + Speaker Diarization + LLM extraction) using RabbitMQ, Redis & LangChain
    • Architected a smart security camera system with real-time facial recognition & object detection (YOLO11 + DeepSort + MTCNN)
    • Explored Multi-Agent Systems with Google A2A & MCP — specialized agents orchestrated by a Host/Manager Agent
    • Applied GCN (Graph Convolution Networks) to drug discovery for biological activity prediction
    • Engineered Vietnamese Hate Speech Detection with PhoBERT & LSTM
    • Performed K-Means clustering and comparative analysis across 10+ ML models (SVM, Random Forest, AdaBoost) and Keras Deep Learning, achieving 94.74% accuracy in weather forecasting.
  • My edge: I don't just train models — I design the full system: async pipelines, containerized services, scalable APIs, and agent orchestration. I care about performance, reusability, and real-world impact.

  • 🌱 Currently exploring: Context Engineering · Advanced RAG · Multi-Agent Orchestration · MCP Protocol

  • 👨‍💻 All my projects: github.com/HuNguyen51

  • 💬 Ask me about: LLMs · Computer Vision · Multi-Agent Systems · AI System Design

  • 📫 Reach me at: hunguyen3525@gmail.com


🚀 Core Stack:

Languages & Runtime

python nodejs

AI / ML Frameworks

pytorch tensorflow scikit_learn opencv

LLM / GenAI Ecosystem

langchain ollama  LLMs  RAG  ReAct Agent  Multi-Agent  MCP  Prompt Engineering

Computer Vision

 YOLO11  DeepSort  MTCNN

Speech

 Whisper  Pyannote

Backend & APIs

fastapi flask

Infrastructure & Messaging

docker redis rabbitmq git

Databases

mongodb mysql  Vector DB


🧩 Also Familiar With:

pandas seaborn javascript html5 css3 bootstrap linux postman firebase kotlin android arduino  MQTT  Selenium  GCN  PhoBERT


hunguyen51

 hunguyen51

hunguyen51


Popular repositories Loading

  1. Air-Quality-Forecasting-Challenge Air-Quality-Forecasting-Challenge Public

    Dự án xây dựng hệ thống dự báo chỉ số chất lượng không khí dựa trên dữ liệu lịch sử và các yếu tố môi trường như nhiệt độ, độ ẩm, tốc độ gió bằng các kỹ thuật Machine Learning.

    Jupyter Notebook 1

  2. HuNguyen51 HuNguyen51 Public

    1

  3. machine-learning machine-learning Public

    Dự án thực hiện phân tích toàn diện bộ dữ liệu thời tiết (2015–2021) thông qua kỹ thuật phân cụm K-Means và so sánh hiệu suất của hơn 10 mô hình học máy khác nhau.

    Jupyter Notebook 1

  4. Transformer-Encoder-Text-Classification Transformer-Encoder-Text-Classification Public

    Dự án thực hiện phân loại văn bản báo chí từ bộ dữ liệu VNTC bằng cách sử dụng kiến trúc Transformer Encoder. Quy trình bao gồm tiền xử lý tiếng Việt với Pyvi, mã hóa văn bản qua mô hình pre-traine…

    Jupyter Notebook 1

  5. image-classification image-classification Public

    Dự án phát triển hệ thống Multilabel Image Classification trên bộ dữ liệu MIML, sử dụng các kiến trúc Deep Learning: AlexNet, VGG16 và ResNet18. Hệ thống cho phép gán đồng thời nhiều nhãn (biển, nú…

    Jupyter Notebook 1

  6. security-camera security-camera Public

    Hệ thống camera an ninh thông minh tích hợp Computer Vision và Deep Learning. Dự án sử dụng MTCNN và FaceNet để nhận dạng khuôn mặt, kết hợp YOLO và DeepSort để phát hiện, theo dõi đối tượng đa ngu…

    Python 1