Skip to content
View franktsaodev's full-sized avatar

Block or report franktsaodev

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don鈥檛 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鈥檚 behavior. Learn more about reporting abuse.

Report abuse
franktsaodev/README.md

Hi, I'm Frank Tsao 馃憢

AI Application Engineer focused on Python, Computer Vision, AI Agents, and Automation.

I build AI applications and software systems with experience in computer vision, OCR, automated testing, LLM-based agents, and production-oriented software development.

My work spans from computer vision and 3D LiDAR research to AI agent architecture and backend engineering.

馃殌 Featured Projects

Frank AI Agent

A modular AI agent framework built with Python and FastAPI for developing stateful, extensible, and observable LLM applications.

Highlights: Agent Sessions 路 Conversation & Fact Memory 路 Tool Calling 路 Plugin Architecture 路 Structured Tracing 路 FastAPI 路 Docker

3D LiDAR Pedestrian Detection and Tracking System

A real-time pedestrian detection and multi-target tracking system developed as my master's thesis at National Chung Cheng University.

The system processes point clouds from a Velodyne VLP-16 and combines PCL-based point cloud processing, AdaBoost classification, Kalman Filter + NNDA tracking, and DBSCAN grouping for pedestrian detection and social-distance risk analysis.

Highlights: 3D LiDAR 路 ROS 路 PCL 路 AdaBoost 路 Kalman Filter 路 NNDA 路 DBSCAN 路 Gazebo 路 RViz

Traditional Chinese Scene Text Detection

A computer vision project developed for a Traditional Chinese Scene Text Recognition Competition, focusing on the text detection stage.

The detection pipeline is based on Differentiable Binarization (DB) with a deformable ResNet50 backbone, multi-scale feature fusion, and OpenCV-based post-processing for text bounding-box extraction.

Highlights: Computer Vision 路 Deep Learning 路 OpenCV 路 Differentiable Binarization 路 ResNet50 路 Scene Text Detection

馃洜 Tech Stack

Languages

Python 路 C# 路 C++

AI & Computer Vision

OpenCV 路 YOLO 路 OCR 路 Computer Vision 路 3D LiDAR 路 Point Cloud Processing

AI Agent Engineering

LLM Applications 路 Tool Calling 路 Memory Systems 路 Plugin Architecture 路 Prompt Engineering 路 Structured Tracing

Backend & Engineering

FastAPI 路 REST APIs 路 Docker 路 Git 路 Pytest 路 Ruff 路 Pyright

Research & Robotics

ROS 路 PCL 路 Gazebo 路 RViz 路 AdaBoost 路 Kalman Filter 路 DBSCAN

馃幆 Current Focus

  • AI Agent architecture and LLM application engineering
  • Computer Vision and AI automation
  • Production-oriented Python development
  • Building reliable and maintainable AI systems

Pinned Loading

  1. Frank-AI-Agent Frank-AI-Agent Public

    A modular AI agent framework with memory, tool calling, plugin architecture, tracing, session management, FastAPI, and Docker.

    Python

  2. people_tracking people_tracking Public

    Real-time pedestrian detection and multi-target tracking using 3D LiDAR, AdaBoost, ROS, PCL, Kalman Filter, and DBSCAN.

    Makefile 1

  3. scene_text_recognition scene_text_recognition Public

    Traditional Chinese scene text detection using Differentiable Binarization and OpenCV.

    Python