Global Security - Security Control VerificationGlobal Security - Data Protection

Data platform • Big data ETL & tuning – Built highly efficient Apache Spark ETL jobs that handled data of daily incremental size of more than 800TB uncompressed; optimized under-performing tasks & reducing runtime from 3 hours+ to around 25 minutes. • System design – Architected & executed (full-stack) a self-service machine learning tool using FaaS (Function as a Service), ClickHouse, Kafka, Flink, and in-house visualization tools, among others, enabling non-technical users to evaluate revenue performance at scale and saving iteration time from 20 minutes to a few seconds. • Stakeholder management – Interfaced with stakeholders to understand data need and partnered to develop metrics for livestreaming business; earned trust and improved team’s NPS from 20% to 60%.

Data platform

• Data modeling – Architected and led the development of a Snowflake data warehouse, using data vault & dimensional modelling techniques and DBT; delivered product significantly more flexible (reduced new feature development time from 4 days to 1 day) and maintainable (ability to modify logic in one place thanks to modularization & code reuse, compared to 5 – 10 places before).• Data pipelines – Built both schedule-triggered and event-driven batch data pipelines using Airflow, SSIS, and PowerShell with average uptime > 99% (SLA being 75%); established alerting mechanisms achieving incident mean time to resolve < 1 hr (used to be “until noticed by user” + 1 day).• Streaming – Implemented a streaming analytics solution to understand legacy SQL Server usage, using extended events, C# (.NET Core), Kafka, and Flink, in order to make data-driven design optimizations for the above data warehousing project (e.g. level of denormalization); project was the first within department to measure and quantify system usage and enable informed decision-making.• Grow the team – Mentored new developers and authored department’s first-ever “labs” for hands-on practice, introducing “self-service” onboarding and reducing “time to readiness” from 1 week to 2 days; Presented Airflow, Kafka, and DBT to company-wide audience, of which the Airflow presentation was rated best of the year and the tool was adopted by other departments for automation tasks.• Project leadership – Identified opportunity to automate mentorship matching offered by ERGs and led a group of 18 developers to build the first-ever matchmaking platform company-wide; project is work in progress which already attracted interest from 3 ERGs, and by completion will transform current manual, week-long process into instantaneous, optimal matching.

**Recipient of Impact Award (2 students within the whole school)** - Utilized Spark on Databricks platform, ran FP-Growth algorithm to mine association rules and analyze halo products- Authored an object-oriented Python package of reusable building blocks, in order that the company can build upon this project in the future. The package was built with special optimization for performance and cloud-ready.- Set up Azure DevOps continuous integration (CI) pipelines to automatically build the aforementioned package and its Sphinx documentation; maintained and managed Git repository for the whole team.The project is provided through Washington University Center for Experiential Learning. This is not full time employment at Direct Supply.

Teaching assistant for CSE247 Data Structure and Algorithms

Teaching Assistant for SQL and Database Management

• Worked closely with business partners and trained an XGBoost model for backorder prediction; addressed data leakage issue with special cross validation technique; presented actionable result to C-suite executive and other non-technical business partners• Working closely with IT, architected and provisioned the company’s first dedicated lab environment for data science, consisting of cloud-based database and auto-scaled compute resources, among others, achieving time saving of 70% (from 5.5 hours to 2 hours, per iteration); this was beyond responsibilities assigned by manager• Launched a comprehensive documentation SharePoint site for efficient and scalable knowledge transfer; site was the first of its kind organization-wide and received near 500 views; this was beyond responsibilities assigned by manager

• Executed data analysis with Python, SQL - Teradata, Hive, and Spark for shipping from Japan to the US and optimized delivery estimate accordingly; the project had impact over 4 million transactions every year and results were presented before Asia-Pacific department head• Reviewed defect (bad buyer experience) performance of Greater China sellers with Tableau and SQL; ran regression analysis for forecast and proposed policy recommendations for regional lead’s consideration• Automated routine processes (Business Intelligence reporting) on Linux, saving the team at least 5 hours per week• Authored New Intern Guide that addresses pinch points for office newbies, which was later adopted by various teams; drafted detailed tech tutorial and received more than 10 stars on Company GitHub• Earned reputation of the "go-to" person in office for SQL query optimization and debugging

• Implemented company’s first automatic fraud prevention system by modeling past transactions; frauds detected were three times more than previous experience-based manual processing• Introduced procedures that significantly increased ETL system efficiency by utilizing big memory advantage; shortened delivery time of data requests by at least 50% and made possible data analysis of TBs of weblog