As one of the most senior members of the backend team, leading the rearchitecture of Deliveroo Plus ecosystem integration under DoorDash platform with 3 other Senior/Staff engineers.
This involves complete decommission of 9+ Deliveroo Plus backend services from AWS and rewriting in a greenfield "infinitely scalable" Lambda and Cassandra tech stack under generalised DoorDash/Wolt/Deliveroo umbrella.
In the Plus team, delivered a new "Diamond" premium subscription tier for UK customers with GraphQL-server driven UI, Click-to-Claim On-Time-Promise experience serving 700 RPS and processing 3K claims a day. Led the migration of legacy Ruby-based flow claims to the new stack with gRPC and DynamoDB datastore.
Introduced annual subscription billing for EU & UK markets, which was adopted by 6K customers in the first month; led the implementation of idempotent payment workflows for the high-cost transaction tier switches.
Built a restaurant recommendation engine in Go that used an OpenAI LLM to understand and act on user prompts.
Owned reviews & ratings microservices fleet (Go + PostgreSQL + Snowflake), storing 200M+ reviews annually and serving 1K RPS. Built a Kafka event stream and a DynamoDB-backed Content Moderation service in Go to interact with third-party moderation APIs.
Owned and maintained a tier-0 Order History Go microservice with PostgreSQL database serving 3K RPS. By redesigning queries, table structures, and index conditions, reduced the average latency by 2x and eliminated 99.99% of timeout errors. Reduced the AWS Database cost by 40%.
Led a team of four to build a Customer Support web application with Java Spring to answer customer requests and examine orders in a single unified application.
Built a new dockerised microservice in Java Spring for the Fraud team to control the lifecycle of order actions. Used ActiveMQ message broker to provide async processing of order actions over MS SQL Server database.
In the Order Pipeline team, developed and maintained Java Spring microservices for an automated flow of the order progression with an event-driven architecture and best warehouse selection optimisation problems.