Jeremy Hsu

Professional Experience

June 2023 - Present · Seattle, WA

May 2022 - August 2022 · Menlo Park, CA

Implemented the backend for 3D avatar sticker categories on platforms such as the Facebook news feed using Hack (PHP) and GraphQL, driving increased user engagement with avatars and helping them find their desired sticker over 23% faster.
Designed and built a new internal tool using Hack, React, and Relay that enables engineers, cross-functional partners, and graphic artists to directly manage avatar stickers and categories.
Refactored client GraphQL queries to use the new 3D sticker category backend architecture and wrote a migration script in Hack to create new 3D sticker categories based on existing 2D sticker categories.

May 2021 - August 2021 · San Francisco, CA

Architected and developed the end-to-end refund flow, user lifecycle API, and an endpoint to fetch real-time vehicle data through QR code scanning or license plate entry for a server-side API build with Ruby on Rails and MySQL, enabling seamless MaaS integrations between Lime vehicles and public transit apps.
Worked closely with engineers and cross-functional partners at Lime and external partners to ship the API to production and complete a successful inaugural launch in Berlin in August 2021.
Extended the functionality of a company-wide internal administrative portal by creating React UI components that support assigning permission access-control roles and viewing mobility data feed instructions for regions and external partners.

May 2020 - May 2021 · San Jose, CA

Significantly improved developer productivity by building new Python-based internal tools, deployed using Jenkins or CLI interfaces, to query, analyze, and triage client logs for a cross-platform media SDK for Webex Meetings and Teams.
Implemented a new data aggregation layer in the SDK's metrics ETL pipeline using Python, the Elasticsearch API, and Airflow, increasing Kibana dashboard load times by over 500%.
Created and monitored interactive Kibana dashboards visualizing critical media quality and network performance metrics, enabling Webex engineers to rapidly identify problematic trends and issues.

February 2022 - Present · Cambridge, MA

Advised by Dr. Stratos Idreos on research projects related to self-designing data systems and systems for machine learning.
Developed a workload-aware, R-tree-based partitioning algorithm that determines the optimal set of specialized indexes across partitions of a dataset, achieving up to a 53% workload execution time reduction over the lab's flagship self-designing system and up to an 11% improvement over Meta's RocksDB.
Built a Python platform for conducting fast and robust HPO scheduler experiments by applying an expressive parametric modeling approach to accurately simulate loss landscapes on real-world datasets.

September 2019 - May 2022 · Cambridge, MA

Datamatch is a student-run matchmaking service working closely with local businesses to connect college students based on their personality compatibility with free meals.
Led a team of 20+ students to redesign and add improvements such as user bio sentiment analysis to Datamatch's scoring and stable machine algorithms written in C++ and Python, which has scaled to match 50,000+ students across 40+ schools.
Improved Datamatch's pairwise compatibility score function by implementing sentence embeddings with Sentence-BERT to analyze user sentiment similarities and dynamic inverse-proportional weightings to address answer distribution polarities.

September 2018 - April 2019 · Mountain View, CA

Handled venue and catering logistics for MVHacks, a 12-hour 150+ person hackathon event that connects talented teams and students to mentors, resources, and workshops to help them realize their potential.
Played an integral role in company outreach and in securing over $4,000 of financial partnerships with Cisco and Ford Greenfield Labs.