🌁Me As A Picture
Contact
Work Experiences
Georgia Institute of Technology
Aug 2026～Dec 2026Incoming Grad Teaching Assistant @ CS 6604 Conversational AI
Atlanta, Georgia
Walmart
Jun 2026～Aug 2026Data Scientist III Intern @ Sam's Club AI Acceleration Team
Bentonville, Arkansas
- Integrated LLM & RAG with filtered vector search into dashboard for answering camera status queries at camera/area/site levels, with cosine similarity & LLMs-as-a-Judge as quality gates against hallucinations.
- Designed an embedding signature to verify Milvus vector database ingestion & retrieval sides call the identical embedding model — despite being forced onto different API clients — eliminating silent retrieval degradation.
- Deployed a monthly Gemini Vision CronJob across 1,245 in-store cameras spanning 9 sites to classify snapshot content and align each camera to its current area, using SVG floor layouts as the primary match reference with CSV data as fallback.
- Achieved full-fleet inference of 1,245 cameras in 13 minutes through concurrency, while capping glibc malloc arenas and pre-downsampling images to eliminate out-of-memory failures.
- Deployed an hourly CronJob pipeline with Docker and Kubernetes, integrating asynchronous processing that reduced 50% execution time for capturing 1,245 camera snapshots into Azure, using 7-day TTL to save 75% storage cost.
SUEZ
May 2022～Sep 2024Data Scientist @ Smart & Environmental Solutions (SES) China
Shanghai, China
- Streamlined an NLP-based Keras classifier for imbalanced SCADA alarms, reaching 82% production accuracy.
- Designed a DBSCAN-based anomaly detector for air compressors predictive maintenance, achieving 70% production F1 score and forecasting failures two weeks in advance despite having only a single feature.
- Accelerated clustering on 12 air compressors with Joblib, reducing runtime by 20% in Linux CronJob.
Pramerica Life Insurance
Sep 2020～Apr 2022Sep～Dec 2020: Intern; Dec 2020～Apr 2022: Data Management Trainee
Shanghai, China
- Built the company's first BI sales data platform into both PC and WeChat mini programs.
- Developed multidimensional filtering, groupby & deduplication functions that saved 50% operation time.
Skills Bag
⛓️ LLMs & Agents
RAG, Hugging FaceSupervised Fine-tuning
Distillation
🪄 Gen AI & ML
🏎️ Automation & CI/CD
☁️ Cloud & Data Platforms
🛢️ Databases & C++
👀 Web & Visualization
Education
Georgia Institute of Technology
Aug 2025～Dec 2026MSc Analytics
@ School of Computational Science and Engineering
Atlanta, Georgia
- GPA 4.0/4.0
Conversational AI, Natural Language, Machine Learning, ML in Computational Biology, Data & Visual Analytics, Computing for Data Analysis, Nonparametric Data Analysis, Business Analytics Practicum, Data Analytics in Business
University of California, San Diego
Sep 2019～Sep 2020Master of Finance @ Rady School of Management
La Jolla, California
National Taiwan University
Sep 2014～Jun 2018Department of Economics @ College of Social Sciences
Taipei, Taiwan