Experience
2026 — Now
2026 — Now
San Francisco Bay Area
▶ Able to design, train, and evaluate machine learning models using Logistic Regression, Random Forest, XGBoost, and Neural Networks for classification and prediction tasks.
▶ Experienced in applying NLP techniques, including BERT-based models, to sentiment analysis problems with datasets of up to 50,000 samples, achieving 95% model accuracy.
▶ Skilled at feature engineering, dimensionality reduction (PCA), data balancing, and comparing multiple modeling approaches to identify optimal solutions.
▶ Proficient in building end-to-end ML workflows, from data preprocessing and model training to evaluation and backtesting for decision-making use cases.
▶ Able to analyze model performance critically, identifying limitations of clustering-based approaches (e.g., K-means) and iterating toward tree-based and ensemble models.
▶ Experienced in backend system design, including building microservices using Go and gRPC for efficient inter-service communication.
▶ Capable of designing scalable data pipelines, integrating MongoDB, Elasticsearch, Redis, and message queues (Google Pub/Sub) to support search and ranking systems.
▶ Able to translate ML outputs into production-ready systems, collaborating ML logic with backend services to improve feature utilization and system resilience.
▶ Strong foundation in software engineering practices, including API design, microservice architecture, containerization (Docker, Kubernetes), and cloud deployment (GCP).
▶ Well-versed in bridging ML and backend engineering, ensuring models are not only accurate but also deployable, maintainable, and scalable in real-world systems.
2025 — 2025
2025 — 2025
San Mateo, CA
• Designed and deployed a stateless LLM-driven agent system that dynamically generates MCP server configurations by processing user requests and conversation history.
• Built an agent orchestration workflow that constructs prompts from chat history and user inputs, enabling step-aware reasoning and iterative task progression across requests.
• Implemented a retrieval-augmented generation (RAG) pipeline using vector embeddings stored in ChromaDB, applying cosine similarity to retrieve top-k relevant templates for grounded LLM inference.
• Engineered an ephemeral inference architecture where each request instantiates a new agent instance, ensuring stateless execution, scalability, and deterministic context reconstruction from client-managed history.
2022 — 2023
2022 — 2023
• Developed, shipped, and operated the company’s first microservices architecture deployed on Google Cloud Platform (GCP), handling loads up to 200 QPS.
• Redesigned a new search feature using MongoDB, synchronized data to Elasticsearch via Monstache, applied Redis caching for load reduction, and used gRPC and Google Pub/Sub for inter-microservice communication, using Go.
• Implemented hourly cron jobs for keyword ranking updates using both algorithmic and ML-based
approaches, and migrated related functionalities to microservices, improving system resilience and
increasing search feature utilization by 40%.
• Produced a livestream page gacha system using MySQL, successfully launched and adopted at scale within the first month.
2017 — 2021
2017 — 2021
Xinzhuang District
• Proposed and led the separation of frontend and backend systems, introducing Vue.js for the frontend and PHP for the backend, and establishing shared APIs used across mobile applications and web platforms.
• Automated payroll calculations, reducing manual processing time by 75% for internal administrative workflows.
• Implemented administrative services for consultants, reducing inquiry response time and streamlining insurance process tracking by over 80%.
• Led development of a subordinate management system for internal supervisors, reducing manual coordination and administrative overhead by over 70%.
Education
University of San Francisco
Master's degree
Tamkang University