Chi-Wei (Tomy) Hsieh

Coursework: Cloud Computing, Intro to Database Systems, Introduction to Computer Systems

Honorable Mention for Graduate Capstone Project: Question Generation Quality Enhancement based on BART

Designed and deployed a Kubernetes-based Cloud IDE for 500+ engineers across multiple regions, enabling on-demand CPU/memory/GPU provisioning and strict data localization.

Optimized GPU usage by ~40% via just-in-time mounting/unmounting, reducing idle costs and boosting utilization.

Secured the monorepo by restricting access to authenticated IDE instances, mitigating exfiltration risks while streamlining developer workflows.

Revamped a large-scale CI/CD pipeline architecture, tripling task capacity and slashing deployment time by 80%.

Developed automated deployment workflows in Go, improving reliability of feature rollouts across multi-cloud environments.

Bolstered cloud security via sidecar containerization, enhancing data isolation and compliance with internal standards.

Engineered and sustained a Django-based full-stack solution to visualize test reports, cutting manual analysis time by 50%.

Accelerated bug detection and resolution by 70% using automated Python validation routines, driving faster release cycles.

Boosted hardware test capacity by 80% through time series models in a FastAPI backend for real-time analytics.

Implemented Presto’s dense layout HyperLogLog for fast cardinality estimation over large datasets.

Developed a thread-safe buffer pool manager with LRU-K replacement and a disk scheduler to optimize memory management and disk I/O efficiency.

Built a concurrent B+Tree supporting efficient search, insertion, deletion, and in-order iteration with proper concurrency control mechanisms.

Implemented high-throughput real-time data pipelines on AWS EMR with Kafka and Samza, ensuring near-zero-latency stream processing.

Optimized driver-matching and personalized ad-targeting algorithms, boosting throughput and lowering cost of operations.

Developed a single source of truth for personal finances system with FastAPI, PostgreSQL, Next.js and Docker Compose, automating invoice retrieval, and streamlining multi-country expense management.

Implemented multi-currency & i18n with UTC-based data storage and Decimal fields for reliable cross-border, multi-account expense tracking.

Ensured high reliability and scalability via containerized architecture, CI/CD pipelines, and Pydantic data validation.