Truong Mai Huy

AI Engineer

Download PDF version Up-to-date version of CV is available at

Personal photo
As an experienced AI Engineer with a strong background in computer science and mathematics, I specialize in developing and implementing machine learning algorithms that deliver measurable business value. With expertise in deep learning, computer vision, and AI model optimization, I excel at designing solutions for complex challenges across diverse industries.

Skills

English
French

Professional Experience

August 2022 - Present

DSoft-JSC, Danang, VN

AI Engineer

  • Developed PTZ ONVIF camera control drivers for surveillance systems.
  • Created algorithms to interpret lane divider positions after camera adjustments.
  • Designed and implemented automatic PTZ camera control using reference images.
  • Built algorithms to detect lane dividers based on vehicle trajectories.
  • Enhanced the FaceLog system with features such as head pose estimation, face quality analysis, anti-spoofing models, and optimization of streaming processes.
  • Trained lightweight Age, Gender, and Mask prediction models optimized for smartphones.
  • Contributed to Facebook’s crawler tool to extract user information.
  • Automated email-related tasks using OpenAI APIs.
  • Built C++ shared libraries to enable AI models to run on edge devices.
  • Completed a six-month onsite project to develop a system for measuring the default probability of applications in the fintech domain, achieving an AUC of 0.8.

Python PyTorch ONNX OpenCV Docker Redis C++ DeepStream GStreamer Scikitlearn LightGBM

November 2021 - July 2022

Freelancer, Danang, VN

AI Engineer

  • Contributed to the EVN-NPT PTC2 project by designing a system to measure tree proximity to powerlines using surveillance cameras.
  • Developed a Unity-based anatomy research application (XAP).

Python PyTorch MongoDB Unity

June 2020 - April 2021

FTetch AI, Danang, VN

AI - Intern

  • Re-implemented computer vision algorithms as part of a five-month training program.
  • Developed tools to visualize AI model learning processes.
  • Improved segmentation models using synthetic data augmentation.

Pytorch Tensorflow OpenCV

July 2019 - October 2019

FPT Software, Danang, VN

Software Engineer - Intern

  • Developed a sales website using Java and SQL.
  • Enhanced algorithmic problem-solving skills using the Codelearn.io platform.

SQL Java

Education

Bachelor of Software Engineering. Danang University of Technology [2016 – 2022]. GPA 3.5

Achievements

Certifications

Additional Experience

Container AI

    • Designed and deployed a computer vision system for monitoring customer behavior in autonomous grocery stores, featuring person-fall detection, item-dropping detection, and suspicious activity logging.
    • Implemented real-time alerting via Telegram and automated video evidence upload to Google Drive.
    • Migrated infrastructure to AWS (EC2, S3, Lambda) to enhance scalability and reliability.
    • Tech Stack: Python, OpenCV, Pytorch, AWS (EC2, S3, Lambda), Google Drive API, Telegram Bot API, Docker.



ChaterBot

    • Built an automated content generation and posting bot using n8n, enabling users to query the bot for real-time market and social media insights.
    • Leveraged n8n’s agent capabilities to integrate and aggregate data from multiple sources including X (formerly Twitter), market APIs, and LLMs.
    • Implemented pipelines that crawl trending posts, use ChatGPT to generate responses or original tweets, and automatically post content.
    • Developed features to fetch top cryptocurrencies by market cap and generate corresponding tweet content.
    • Demo: quangthinh.io
    • Tech Stack: n8n, OpenAI (ChatGPT), X/Twitter API, CoinGecko/CoinMarketCap API, JavaScript, Webhooks, REST APIs.



OpenDrawing

    • Developed a vision-based automation pipeline to support wire drawing processes by detecting and tracing lines from raw material images.
    • Implemented line detection and tracing algorithms to identify and connect structural components in engineering drawings.
    • Enabled automated recognition of drawing components to facilitate digital interpretation and downstream processing.
    • Demo: opendrawing.ai
    • Tech Stack: Python, OpenCV, Scikit-image, Custom Line Tracing Algorithms, NumPy, Computer Vision.



XOCR

    • Built an automated system to extract tweet content from scrolling screen recordings on X (formerly Twitter).
    • Designed a visual pipeline that detects dynamic tweet regions, stitches frames into a full screen map, and segments individual tweets.
    • Applied OCR techniques to extract structured information from tweet snapshots, enabling downstream analysis or automation.
    • Optimized tweet detection using movement-based region extraction and image processing heuristics.
    • Tech Stack: Python, OpenCV, Image Stitching, Tesseract OCR, NumPy, Scikit-image, Computer Vision.



SwingSense

    • Developed a computer vision system to analyze bat swing motion in sports scenarios using video input.
    • Implemented bat tracking algorithms to detect and follow the start and end points of the bat during a swing shot.
    • Built a ball detection pipeline that identifies seam patterns and analyzes rotational behavior during flight.
    • Enabled extraction of key biomechanical metrics such as swing angle, timing, and spin analysis for coaching insights or performance diagnostics.
    • Demo: ls.latentia.ai
    • Tech Stack: Python, OpenCV, Computer Vision, Pose Estimation, Object Detection (YOLO), Image Segmentation, Optical Flow, Rotation Estimation.



Aviso

    • Built a Selenium-based automation system to perform daily tasks on an online platform.
    • Developed an OpenAI CLIP-powered module to automatically solve hCaptchas.
    • Tech Stack: Python, Selenium, OpenAI CLIP, PyTorch.



Virtual TikToker

    • Automated content creation and publication on TikTok, targeting viral trends to grow followers.
    • Designed video generation algorithms using computer vision and machine learning techniques.



XAP

    • Created an interactive 3D anatomy learning application with real-time collaboration capabilities.



IOC

    • Migrated a Vue-based IoT resource monitoring interface to Angular, enabling streamlined notifications and coordination for Danang City.



Categories: Python Javascript React Angular Vue Unity Docker