AI Engineer in Taiwan · Data Scientist at Harbor Tech
I turn industrial signal analysis into products people can operate: equipment monitoring, diagnostic reporting, and end-of-line quality inspection. My work spans product and domain design, system integration, customer delivery, and the development environment that helps people and AI agents work together.
I lead product and domain design, connecting vibration and partial-discharge (PRPD) analysis, equipment status, diagnostics, scheduling, reports, and operator workflows. With the team and AI agents, I built much of 2.0's functionality from scratch while handling coexistence and gradual migration from 1.0. As of October 2026, 2.0 operates independently of 1.0.
I led the customer project from requirements and specifications through system implementation, UI/UX refinement, and on-site demonstration and training, working with AI agents and engineering partners. When the initial UI met specifications but required deep, circular navigation, I reorganized it around plan-list, work-order, and board views, and added sensor-installation checks before measurement.
I built traceable project and issue records linking problems, decisions, implementation, and verification. AI agents develop, repair, and verify in parallel on isolated branches; daily functional regression feeds the maintenance loop. I set quality criteria, review domain assumptions, and work with reviewers on merge and release decisions.
My master's thesis and most complete research project: research design, multi-scale CNN architecture, data-pipeline integration, ablation studies, cross-dataset evaluation, and explainability analysis. Some preprocessing was adapted from ChINN.
A team project combining tracking, person re-identification, and face recognition. I primarily owned ReID integration, sampling and matching, similarity thresholds, and tracking-record orchestration, using FastReID as the base model. The team won first place in Soochow University's capstone competition. Demo
I proposed the topic, led team coordination, and designed, implemented, and deployed the Telegram chatbot that integrated teammates' recommendation modules. Our team received an Excellence Award in the 2020 Financial Programming Course Competition.
I contributed topic ideation and data collection, using a friend's dashboard-template fork with the team. The team advanced to the final.
- Knowledge-Based Systems, 2024: Dynamic traffic network representation model for improving the prediction performance of passenger flow. Coauthor; contributed supporting code and technical assistance. Paper
- IIAI AAI, 2021: A Dynamic Embedding Method for Passenger Flow Estimation. Contributed method ideation, data processing, model construction, and experiments. Competitive Paper Award. Paper
- TCSE, 2020: Hierarchical Attention Network with Sentiment Lexicons for Stock Article Emotion Analysis. Conducted the research for this coauthored study. Paper
Python · Django · PostgreSQL · Redis · Docker · PyTorch · scikit-learn · NumPy · pandas · SciPy · React · TypeScript · Playwright
Additional team award: Creative Award, AI for Asset Activation Innovation Contest, 2022.