Nguyen Huu DungAI Engineer
Profile

Innovative AI Engineer with hands-on expertise in LLMs, scalable RAG frameworks, and multi-agent systems. Proven ability to integrate modern AI stacks to build production-ready, autonomous applications that drive operational efficiency.

EDUCATION

Posts and Telecommunications Institute of Technology

Bachelor of Information Technology
2021 – 2026
  • •Merit-based Scholarship Recipient (4 times)
    • Professional Experience

    LLM Engineer - Aimesoft

    11/2025 – 05/2026

    Backend / AI Engineer | AIMeTrans – AI-Powered Translation System

    Developed a comprehensive, session-based document translation and interactive chatbot platform supporting multi-format files and domain-specific customization.

  • •Built a high-performance backend using Python (FastAPI, AsyncIO) and LangChain/​LangGraph, integrating OpenAI, Claude, and Azure for contextual translation.
  • •Designed smart batching and prompt thinning that reduced token costs by 65% and slashed latency by 75% (from 12.8s to 3.2s).
  • •Developed custom parsers (PDF, DOCX, PPTX, XLSX) that maintain 100% of the original document formatting post-translation.
  • •Implemented a session-based workflow allowing real-time translation adjustments via an AI chat agent.
  • •Integrated embedding vector search, custom domain dictionaries, and containerized the system using Docker.
  • •Participated in the Sign Language Recognition Project using skeleton-based sequence data.
  • •Improved model accuracy (~1.0 across 120 classes) by implementing SiFormer – Feature-Isolated Transformer for skeleton-based sign language recognition.
  • •Researched and experimented with TensorFlow.js for deploying AI models on web browsers.
  • •Optimized data preprocessing workflows: normalization, affine transformation, and landmark alignment to improve recognition stability.
    • PROJECTS

    AI Chatbot for Online Medical Appointment Scheduling⁠

    Python, Rasa, RAG (FAISS, BM25), MCP, Streamlit, SQL
    2025
  • •Developed an end-to-end AI medical assistant capable of automating appointment scheduling, retrieving physician data, and delivering symptom-based specialty recommendations.
  • •Architected a complete NLU-DM-NLG pipeline utilizing Rasa, training a DIET classifier model to achieve an 88.9% accuracy rate across all intent classifications and entity extractions.
  • •Integrated a Retrieval-Augmented Generation (RAG) framework using FAISS and BM25 to enhance dynamic knowledge retrieval and improve the accuracy of symptom-to-specialty matching.
  • •Implemented the Model Context Protocol (MCP) to seamlessly manage multi-turn conversational context, resource routing, and complex tool coordination.
  • •Built an interactive Streamlit user interface integrated with an SQL medical database to deploy and demonstrate a fully functional AI pipeline.
    • SKILLS
    Language: Python, Java, C++, JavaScript
    Tools: Google Colab, Jupyter Notebook, Git, Docker
    Databases: SQL Server, Firebase Realtime Database
    Frameworks & Libraries: PyTorch, TensorFlow, Pandas, scikit-learn, NumPy, Matplotlib, OpenCV, LangChain, LangGraph
    English: Intermediate
    LEADERSHIP

    Organization: IT Faculty Student Union, PTIT

    Executive Committee Member
    2022 – 2024
  • •Planning, executing and organizing events and workshops; coordinating with faculty for smooth operations.
  • •Developed leadership, communication, and organizational skills.