</>
Hi, my name is

Nguyễn Chí Trai

AI Engineer

I'm a software engineer specializing in Computer Vision and End-to-End AI. I architect and deploy robust AI systems, turning complex algorithms into scalable, commercial-grade products.

Check my work Contact me
01.

About

Hello! My journey started when I tried to make a computer "see" and understand the world. Fast-forward to today, I'm building End-to-End AI pipelines that solve real-world problems—now enhanced with Agentic capabilities.

I don't just train models; I build the entire pipeline—including custom visualization tools. I write the frontend code to bring model outputs to life, bridging the gap between complex algorithms and interactive user interfaces.

Core Focus

End-to-End Pipelines, Computer Vision, AI Agents

Approach

Commercial-Ready, Optimized, Practical

Profile
02.

Skills

Computer Vision

YOLO v11 YOLO OBB PaddleOCR OpenCV CLIP/VLMs

Pipeline & Backend

Python FastAPI GenAI Integration Docker REST APIs

Data & Tools

CVAT Fine-tuning Google Colab Dataset Engineering
03.

Work

Agentic AI

AI Agent Controlling Web UI

An interactive Agentic interface transforming natural language into structured UI commands. Currently powered by Gemini Flash, the architecture is model-agnostic, allowing seamless replacement with any capable local LLM for flexible deployment.

  • Gemini Flash
  • Agentic Workflow
  • Structured Outputs
View Interactive Demo

Spatial Computing

Floor Plan Analysis

An intelligent pipeline that analyzes 2D floor plan images to generate structured topological data for 3D visualization. The system combines YOLO-OBB and geometric algorithms to detect walls and openings, enabling automated spatial reasoning and interactive 3D reconstruction.

  • YOLO OBB
  • OpenCV
  • Spatial Reasoning
  • Vectorization
View Interactive Demo

Visual Search

Query-by-Example Search

A high-performance visual search system using vector similarity to retrieve objects by example. It allows users to instantly identify repetitive components in technical drawings without text queries or model retraining, ensuring efficient and scalable workflow automation.

  • Feature Extraction
  • Similarity Matching
  • OpenCV
View Interactive Demo
OCR System

Document Processing

Image to Markdown

A comprehensive pipeline transforming document images into structured Markdown. Orchestrates specialized Vietnamese OCR and table processing models to accurately preserve layout and content, ensuring high-quality output for downstream applications.

  • PaddleOCR
  • Deep Learning
  • Table Analysis
Tech Detail

Algorithmic Vision

Technical Detail Detection

Non-deep-learning solution using pure computer vision algorithms to detect technical symbols like elbows and vents. Extremely fast and lightweight for edge deployment.

  • OpenCV
  • Morphology
  • Geometric Analysis
04. Contact

Get In Touch

I'm actively looking for new opportunities to apply my skills in Computer Vision and AI. Whether you have a question or just want to connect, I'll do my best to get back to you!

Email Me