About Me
Full‑stack & data‑platform engineer with production LLM UX, scalable search/data systems, and quantitative research. Built a real‑time React (TypeScript), FastAPI, and WebSockets chat streaming tokens across LLM providers; engineered equities pipelines with regime clustering that reduced max drawdown by 77.7%; and operated multi‑region Kubernetes services to 99.9% uptime while accelerating Elasticsearch queries by 50% after migrating from MongoDB. Currently pursuing BS in Computer Science at USC.
Education
USC CS
Experience
3+ Years
Publications
2 Papers
Languages
EN, ZH
Professional Journey
Quantitative Analyst Intern
Draco Evolution • Taipei, Taiwan
- Built an intraday research stack streaming tick/minute data from Polygon.io and executing paper trades on Alpaca for 5‑second signals
- Engineered a Python ML pipeline (StandardScaler, SelectKBest[f_regression], HistGradientBoostingRegressor) with PCA and K‑Medoids regime clustering
- Designed leak‑free, walk‑forward backtests tracking P&L, win rate, cumulative P&L, and max drawdown; achieved 77.7% lower max drawdown vs buy‑and‑hold
- Ensured market‑aware handling by aligning to exchange sessions and maintaining incrementally updated historical datasets from Polygon REST
Software Engineer Intern
TSMC • Taichung, Taiwan
- Operated fab‑support services on Kubernetes across Taiwan, Germany, and Japan with Prometheus/Grafana instrumentation, delivering 99.9% uptime
- Migrated ticketing data from MongoDB to Elasticsearch; tuned index/shard layout and mappings for semantic‑style queries and 50% faster search latency
- Extended Java backend for role‑based submission, cross‑department routing, and export features; deployed via kubectl and manifests
Software Developer
Kolo AI • Pasadena, CA
- Built a persona‑driven chat UI in React/TypeScript with streamed Markdown and inline citations; added icons/tooltips and model defaults
- Shipped Google Calendar integration end‑to‑end (OAuth 2.0 PKCE, create/read, webhook‑driven sync) with concurrency‑safe, idempotent handlers
- Implemented usage/cost attribution across Sendbird/Telnyx with Decimal‑safe accounting, timezone‑correct aggregation, and SQL/Grafana dashboards
- Prototyped a menu‑intelligence data pipeline: multithreaded/recursive crawling, image/PDF extraction, LLM‑assisted cleaning, and APIs
- Delivered core frontend features for mobile/desktop and added moderation profiles (SHAFT) to switch safety modes per session
Research Analyst
University of Cambridge CCAF • Cambridge, UK
- Evaluated digital‑asset policy, environmental footprint (PoW vs PoS), and social‑safeguard considerations for aid programs; translated findings into recommendations
- Co‑authored (2023): Cryptoasset Ecosystem in Latin America and the Caribbean; Considering Digital Assets for Humanitarian Cash‑Based Transfers
Technical Skills
Languages
Frameworks
Infrastructure
Data & ML
Beyond Code
When I'm not coding, you'll find me gaming, taking on escape rooms, or out on the pickleball court. I split my time between the gym and tinkering with my 3D printer—turning ideas into little fixes, fun prints, and the occasional overengineered gadget. I also enjoy watching anime and Oscar films. If you're into any of that, I'm always down to chat.