Architected a high-availability replication system that increased platform uptime from 99.9% to 99.995%, directly supporting a $1.2M increase in annual revenue by minimizing transaction-disrupting outages.
Implemented a Vector Clock mechanism to enforce causal consistency, which eliminated over 98% of data ordering bugs and reduced customer support tickets related to data integrity by 60%.
Engineered a CRDT (Conflict-Free Replicated Data Type) model that automated data merging across geo-distributed nodes, decreasing write latency by 30% and reducing infrastructure costs by 15% through more efficient conflict resolution.
This new architecture enabled the system to scale and handle a 200% increase in concurrent users, unlocking a new tier of enterprise customers and contributing to a 10% growth in market share.
Software Development Engineer II
Amazon
Seattle
08.2022 - Current
Proposed and implemented migration to web sockets in API Gateway to replace existing polling mechanism thereby reducing infrastructure cost by $280k annually.
Designed and implemented a scalable system dealing with 7000TPS to track more than 300,000 packages picked up by amazon daily. Increased operator efficiency in pickups resulting in cost savings of $680k annually in EU region.
Designed, developed and oversaw rollout of a feature in iOS app widely used for pickup of packages across the world used by more than 50k DAU and reduced pickup times from 30 mins to 10 mins where customers enabled bulk pickup.
Designed mechanism to leverage existing systems to implement package replanning system that allows for manual injection of extra-large packages into desired delivery routes, cutting late deliveries of extra large packages by 25%.
Technologies: Swift, Java, Kotlin, Python, Xcode, DynamoDB, ECS, Lambda, S3, SNS, SQS, AWS API Gateway, Web Sockets
Software Engineering Specialist
GE Digital
Bangalore
12.2018 - 01.2021
Proposed and developed a java algorithm for various Spring Boot microservices to port 'cross tenant' functionalities and replace existing mechanism, improving maintainability. Bringing availability from 99.5% to 99.9999%.
Designed and developed a scheduling microservice by leveraging ActiveMQ's internal scheduler to replace Quartz scheduler. Scaled the microservice from scratch to 2000TPS.
Developed Redis cache reconciliation algorithm amongst various spring boot microservices and brought down the maximum time for data reconciliation from 30 minutes to 5 minutes.