Enjoy coding
2021 — Now
San Francisco Bay Area
Architected and Implemented Experiences and Services Host Calendar Platform: Designed and built the unified calendar experience for Airbnb's two key business lines—Experiences and Services—enabling hosts to seamlessly manage availability across multiple listing types with permission checks and multi-listing calendar overview, serving over 27 million calendar events.
Designed and Delivered Extended Availability Window Feature End-to-End: Led full end-to-end design and implementation of the Extended Availability Window feature enabling hosts to manage booking windows beyond the standard time limit, coordinating development across 7+ services with large-scale backfill operations, impacting 5.7 million listings and driving a 1.03% increase in Airbnb GBV.
Developed and Constructed the Host Smart Pricing ML Infrastructure: Engineered the online scoring infrastructure for Host Smart Pricing, delivering real-time nightly price recommendations to a user base exceeding 9 million hosts, consolidating legacy models into a unified TensorFlow architecture while simplifying the serving stack and data pipelines. Employed Java, Python, Scala, and TensorFlow for the implementation.
Engineered and Implemented Feature Data Pipelines for ML Platform: Orchestrated and executed data pipelines in Airflow to enhance the feature set of the Airbnb ML platform, building capabilities for managing cleaning fee analysis, tracking inflation rate adjustments, processing geo-based event data, and supporting pricing model training with reliable feature delivery. Employed Python, Airflow, Spark, and SQL for the implementation.
Developed End-to-End In-House Feature Analysis Tool: Designed and implemented a comprehensive feature analysis tool for Airbnb catering to all data tables across the organization.
Palo Alto, California, United States
Designed and Executed the Uber Hadoop Resource Management System: Crafted and executed the Uber Hadoop resource management system comprehensively, leveraging Java, React, and GRPC. This extensive project involved gathering user inputs, architecting both the user interface and backend, development, presentation, thorough bug testing, user manual creation, and post-launch maintenance. The system effectively allocates computing resources throughout the organization, streamlining operations, and optimizing resource utilization.
Engineered the Yarn Resource Monitor Daemon: Devised the Yarn Resource Monitor Daemon, a pivotal component that substantially enhanced Yarn utilization through more efficient resource management at the host level. This optimization translated to improved resource allocation, ultimately resulting in heightened system performance and cost-effectiveness.
Enhanced Security for Uber's On-Premise Hadoop Yarn Clusters: Implemented security measures to safeguard Uber's on-premise Hadoop Yarn clusters, ensuring data security and compliance with industry standards.
Palo Alto, California
Constructed Uber's In-House Planning Platform's Frontend: Developed Uber's in-house planning platform's frontend from the ground up using React and Redux. I was responsible for the entire project lifecycle, including design, development, rigorous testing, and final delivery, providing continuous support. This innovative platform revolutionized planning processes, enhancing efficiency and informed decision-making.
Using Nodejs, redux to built the Metric-Insights platform to help finance users understand how spending affects downstream metrics.
Using Nodejs, redux to built the User-Insights platform to visualize the customer value.
2018 — 2018
Designed a well-received model concerning a food dataset (25 categories, 40K items) with bounding box from scratch, leading to a decrease in human workload within a fast-paced, high-stress work environment.
Generated custom CNN model to detect complex hand gestures reliably, delivering loss of 0.4 with milliseconds inference time, exceeding expected performance goals by 60%.
2017 — 2017
Greater San Diego Area
Championed an IOS application based on React Native, taking pictures and adding comments to track skin acne while integrating multiple analytic techniques to ensure application organization and usability.
Built custom web application based on Flask to monitor backend status and manipulate data, supporting the uncovering and integration of various insights to improve app efficiencies.
Guided flawless integration of Google Cloud API to store pictures and reduce related operational costs.
Education
2016 — 2018
UC San Diego
Master's degree
2016 — 2018
2012 — 2016
Beijing University of Posts and Telecommunications
Bachelor's degree
2012 — 2016
2012 — 2016
Queen Mary University of London
Bachelor's degree
2012 — 2016