A/B Setup Team

• Implemented caller-based rate limiting in a high-throughput Java-based gRPC Exposure Service on GCP handling of QPS, enforcing per-caller quotas and segment-level throttling that prevented multiple production outages and stabilized system performance during peak traffic.

• Designed event publishing pipelines using AWS Lambda and S3 for production event delivery, and SNS/SQS for study metadata synchronization, enabling reliable async data flow across the A/B testing platform.

• Built the Python-based orchestration layer for the A/B testing platform, serving Snap’s machine learning teams for experiment setup across ads and content recommendation.

• Designed and launched the Holdout Study feature end-to-end: introduced a new experiment type with multi-feature holdouts and automated configuration management to protect control groups.

• Improved reliability by fixing cache race conditions with atomic operations and replacing full DynamoDB table scans with targeted queries, reducing p99 latency and read capacity consumption.

Advanced Shipping Notice Team

• Designed and implemented new REST APIs to replace legacy shipping notice endpoints, extending the data model to support new business requirements while maintaining backward compatibility.

• Led the full data migration lifecycle for shipping notice data from a monolith to microservices. Defined feature scope, architected the migration strategy, and executed end-to-end with zero data loss.

• Drove rollout planning for the shipping notice migration, coordinating with product, QA, and infrastructure teams on validation criteria, staged cutover sequencing, and rollback procedures.

• Built ETL pipelines to migrate shipping notice data to SAP HANA with idempotent writes and batch checkpointing to guarantee consistency during re-runs.

Profile Team

• Redesigned the audit logging service as a standalone microservice with Kafka-based event streaming. This eliminated audit log loss that occurred under peak load in the prior synchronous design.

• Built a logging middleware layer to unify structured logging and request tracing across the team’s REST APIs, replacing inconsistent per-service logging with a single traceable format.

Backend:

• Developed a web application by using Spring framework and Restful APIs on automating the collection of item data and scaling the data collection from 80 categories of items to 6,000+ to get the most granular level of Item Data possible, reducing the needs for hard-coding and the amount of time it takes for Walmart to launch new item data offerings to market.

• Wrote scripts to automatically import data from old system into new application by filtering out useless metadata fields, and export data in several formats, such as JSON, Excel.

• Delivered a metadata management to reduce fields by 85% to 24, and made the tool user friendly.

• Created JUnit tests for all new logic added to the product services to increase service reliability and stability.

• Achieved a dramatic increase in the quality of Walmart’s Item Data, driving a positive customer experience and sales for Walmart as a retailer, and an increase of the user base of the tool from ~50 users to over 250 users, made a number of additional enhancements to the platform infrastructure.

Frontend:

• Implemented a web using React, Redux to visualize attributes data for internal uses and regularly upgraded to ensure web security and system usability.

• Created extra filters to provide more customized visualization on the different fields.

• Optimized users’ experience by adding autocomplete components to search related field data fetched by calling back-end APIs.