Buy with Prime (BwP)

• Led the Photo on Delivery project, reducing refund requests by 20%; Directed cross-functional collaboration with security, privacy, legal, and fulfillment teams to ensure alignment and seamless execution (Java)

• Led the Enhanced Order Timeline project in Customer Support domain, simplified service architecture and improved CS agent efficiency; Mentored two junior SDEs and collaborated with Tracking team to drive successful outcomes (Kotlin)

• Led the integration of Inventory Availability Service into BwP ordering process, enabling item availability validation; Collaborated with Ordering team and mentored two junior SDEs to drive successful delivery (Kotlin)

• Integrated BwP Identity service with SDO AztecService for Login with Amazon functionality (Java)

Fulfillment by Amazon (FBA) Inbound

• Enhanced the intent-to-instock (i2i) system, a backend middle layer for Send To Amazon platform and an orchestration framework to facilitate communication with downstream services (Java, Typescript)

• Led the Single SKU Palletization project to optimize inbound process for pre-packed pallets, benefiting 6% of FBA sellers

• Improved system scalability to increase inbound limit to support up to 5,000 containers per workflow

• Developed functions of data processing, optimization and visualization pipelines (Python); Owned visualization packages

• Generated visualizations for data validation and model evaluation; Reviewed costing and retention models performance

• Built optimization and backfitting pipeline for Intact Ontario insurance pricing algorithms using in-house libraries

• Developed a communication and speech training system for patients under nonverbal conditions, and a companion app for therapists to collect data, view reports and monitor therapy progress

• Led a dev team of 3 engineers; Collaborated with therapists, test users and designers on feedback and updates

• Designed REST API for user profile/images/library/language trees/communication records/reports

• Implemented backend functions in Python and deployed it using Serverless framework on AWS

• Built data pipeline to process user data and generate progress reports

• Optimized big data processing algorithms (Scala, Spark) to process SWIFT transaction records, reducing processing time from 24 hours to 2 hours and significantly enhancing efficiency and throughput

• Developed data pipelines (Spark, Scala, Python) to load, transform and integrate billions of data records from different sources into Hive tables and build a unified financial network

• Implemented entity resolution algorithms to identify the same entities across different data channels

• Developed a web application prototype, leveraging NLP models for text classification, concept extraction and text generation, to accelerate technical document compositions

• Implemented NLP algorithms and trained customized models, using NLTK, SpaCy, Scikit-Learn and Keras

• Implemented frontend and backend functions using Vue, Flask and MongoDB; Created the docker images and deployed the app on an AWS EC2 server