Bringing AI from technology to products that work.
I'm Candice, an AI product manager at IndustrialNext. I started in vision and software engineering. Today, I turn technical possibilities into products that customers can use and teams can keep building on.
I'm interested in what AI changes on the factory floor: helping a new operator learn faster, reducing repetitive setup work for engineers, and making systems easier to deploy.
My work usually starts with understanding the workflow, defining the product scope and acceptance criteria, then working with AI, software, hardware, and customer teams to validate the solution. A question I keep coming back to: what can we reuse beyond this one project?
Industrial AI
Computer vision
Robotics
0 → 1 product development
My technical background+
I've worked on frontend and backend software, as well as camera, edge-device, and field-system integration. At university, I also led a lab team, organizing training and project work. Those experiences taught me to think about user workflows and implementation details together.
I led the product from concept to deployment, combining vision, step-by-step guidance, action verification, and traceability to help operators follow production standards consistently.
2 weeks → 3 daysOperator onboarding time in the project
AI SOP · Real-time action verificationThe system recognizes the operator's action, compares it with the defined standard, and displays the result in real time · 23 secMy role & outcomes+
I defined the MVP, product architecture, and acceptance metrics; coordinated vision, edge computing, software, and equipment integration; and established deployment checklists and exception handling for MES/ERP-connected workflows.
Additional project outcomes included a 60% improvement in inspection consistency and a 50% reduction in rework. These figures describe the results of that implementation.
Equipment setup and changeover often depend on experienced engineers repeating manual adjustments. I map the real workflow and define where AI assistance can help with vision configuration, calibration, robot teaching, and recipes.
THE QUESTION
How can we reduce the engineering effort repeated on every machine?
My role & current direction+
I drive customer requirement decomposition, field workflow analysis, AI opportunity assessment, initial scope definition, and technical alignment between engineering and customers. The focus is turning repetitive vision and robot setup tasks into verifiable, reusable capabilities.
The next step is to validate candidate calibration, point generation, and recipe capabilities: which tasks can be automated, where engineers remain involved, and how much setup effort can be saved.
YUANTOO · PROJECT LEADField validated
AI Vision for Parts Verification
For automotive warehouses with similar-looking parts and limited samples, I designed a PDA workflow from image registration and AI matching to exception review and backend confirmation, including offline caching and human review.
2–5 imagesTo register a new part
My role & approach+
I defined few-shot registration, segmentation and feature matching, multiple verification checks, and exception routing, with recognition response below 2.5 seconds. I also specified PDA-to-backend APIs, workflow states, and error handling for field use.
LIGHTMATRIX 3D · CORE ENGINEERINGLive deployments
Real-time 3D Free-viewpoint Video
As a core engineering and delivery contributor, I worked on multi-camera capture, 4K stitching, GPU rendering, and 5G transmission for live sports and performances. It shaped how I think about the gap between technical performance and reliable delivery.
16 4K camerasReal-time stitching · end-to-end latency under 200 ms
I developed capture and stitching modules with Python, OpenCV, and OpenGL, and contributed to C++/CUDA rendering optimization. I also worked with telecom and hardware teams on requirements, system integration, and live deployment for baseball, basketball, and theater events.
Equipment setup at a basketball eventMore engineering work3D scanning · ball tracking · AI Box+
Software development projects I contributed to at Lightmatrix 3D, 2020–2023.
Automated 3D scanning & reconstruction
I contributed to software development and hardware integration for a 3D scanning system, connecting camera capture, stitching, and model generation. The setup and a reconstructed portrait are shown below.
Multi-camera scanning setup3D model demo · 5 sec
Automated baseball trajectory tracking
Contributed to software development for a baseball trajectory-capture system designed for post-game analysis.
Post-game analysis demo · 5 sec
Interactive AI Box system
Contributed to software development for an interactive AI Box system. These two clips show the on-site user experience.
Industrial AI product planning, AI Setup definition, and customer alignment. Working with overseas engineering teams to develop reusable capabilities across vision, robotics, and industrial software.
JUL 2024 — JUN 2026
YuanToo Artificial Intelligence
Product R&D Director
Led AI industrial SOP development from discovery and architecture to field delivery. Work included edge vision, automotive parts verification, and RFID traceability for tissue culture, with shared product and deployment standards.
DEC 2023 — JUN 2024
Taiwan United Information
System Analyst · Internship
Requirements analysis, backend API development, and validation for e-commerce systems. Defined business rules, interface dependencies, and exception flows using PHP/Laravel and Docker.
SEP 2020 — DEC 2023
Taiwan Lightmatrix 3D Technology
Full-Stack Engineer
Developed real-time 3D vision, free-viewpoint video, and scanning systems. Integrated cameras, edge devices, algorithms, and the software-hardware workflows needed for field delivery.
04 /
Education & research
2022 — 2024
Tamkang University
M.S. in Computer Science
Research on depth-based 3D motion assessment, using Tai Chi as an application to explore how visual perception can support understandable movement evaluation.
2018 — 2022
National Taiwan University of Science and Technology
From a shopping website to small games, these university projects were my starting point for learning web development, interface behavior, and program logic.
Shopping WebProduct, member, and cart interfacesC# · Typing practiceDesktop interface and input interactionC++ · MinesweeperC++ · Match-threeC++ · Blackjack
05 /
Let's connect
AI, products, and real-world problems.
I'm happy to talk about industrial AI, computer vision, robotics, and product development. Away from work, I enjoy the outdoors, reading, and singing.
