Jose Castillo

Software Engineer

Uber•  May 2019 - Present

WallE - Restaurant and Dispatch Walls Management System ▪ Leading 1 engineer, architected and developed a scalable wall management system handling over 1000 queries per second and improving operations by 250% by enabling Ops teams to make marketplace adjustments and increase courier/eater ratio. ▪ Built backend components and API using Go and Cassandra to allow Ops efficiently manage the restaurant and dispatch walls. ▪ Developed Java and Go client libraries used by downstream Eats critical systems to fetch geo-replicated cached read models of the walls to provide better scalability and availability with a guaranteed SLA. Terra – Marketplace Health Monitoring System ▪ Leading 2 engineers, architected and developed metrics monitoring system to provide insights to Ops into what other automated systems are doing to balance marketplace health. ▪ Designed and implemented backend components and data models to provide 8 core Eats real-time metrics, ensuring it’s easy to work with and future proof for seamless onboarding of newer metrics using Go, Pinot DB and Redis. ▪ Collaborated with leads of other Terra’s systems on planning and development prioritization to deliver all 3 systems (metrics, levers, and recommendations) required for Alpha release, delivering the project 10% faster than expected.

Software Engineer

Jet•  August 2016 - May 2019

SkyWalker - Item Setup System ▪ Led 3 engineers, architected and implemented system to integrate with the newly designed Item Catalog System at Walmart to be used as source of truth instead of existent’s Jet.com and all other Walmart brands. ▪ Directed and coordinated development into two independent phases that allowed parallelized implementation across multiple engineers and delivered the project 30% faster than expected. ▪ Designed a new Item Setup data model to be used by the whole Jet.com warehouse platform that allowed for a seamless integration with Walmart’s system and minimal downstream changes for the Inventory and Merchant API teams. FC Rank - Fulfillment Center Rank System ▪ Led 2 engineers, architected and developed Jet’s fulfillment center (FC) ranking system to select the best FC to ship from at merchant selection time. Generated savings over $5M/month by increasing accuracy on shipping delivery estimates. ▪ Designed and implemented a NoSQL based event sourcing system to maintain a throughput of 10 million rank update events per minute with a full event history for accounting. ▪ Directed development prioritization and collaborated with technical leads of the Pricing team to design and implement testing strategies, eliminating data discrepancies, divergence, and ensuring accurate final calculations. Reload - Inventory Replenishment System ▪ Architected system to provide inventory replenishment recommendations based on outstanding demand, facilitating timely response to changing demand from emergency orders or shifted priorities. ▪ Reduced inventory inconsistencies by 98% by removing a series of dual/triple-writes operations and implementing idempotent distributed transactions using Sagas. Mjolnir - Returns Workflow System ▪ Redesigned and implemented the Returns workflow to remove coupling out of the main Warehouse Management System monolith while improving consistency with downstream systems and overall availability. ▪ Collaborated with the Order Management team to modify data contracts to encapsulate internal data schema. Favored availability by making this team receive data from Kafka asynchronously. ▪ Leveraged eventual consistency by removing coupling with overall system by creating read models of Totes, LPNs and Damaged locations using Azure DocumentDB.

Software Engineer

Florida International University•  April 2015 - April 2016

- Developed real time notification system enabling users to subscribe to custom alerts using Redis Pub/Sub and WebSockets; capable of sending thousands of notifications per second across ITPA’s servers. - Reduced API response time of bus location service by 85% via state memory caching and query optimization.