SHUBHAM SHARMA

Dynamic Technical Lead with a proven track record at Paytm, driving AI initiatives and enhancing global deployment capabilities. Expert in Core Java and microservices architecture, I excel in performance optimization and mentorship, empowering teams to achieve operational excellence and innovate solutions that scale effectively.

• PARTITIONING MYSQL DB (PAYTM), Introduced the database partitioning in our production running application. This is one of the very critical tasks I did in Paytm with Zero downtime and we have to support refunds of past orders. This helped Paytm to reduce DB costs by 80% of our total DB cost.
• REACTIVE/RX JAVA INTRODUCTION (PAYTM), As part of optimizing a critical business flow reliant on a risk service with high latency, I introduced reactive Java concepts using Spring's reactive HTTP client. By unblocking threads to handle incoming requests while waiting for responses, significantly reduced CPU utilization from 65% to 40%. This optimization enabled us to decrease the number of pods in production, resulting in substantial cost savings for the organization.

• PAYMENT-CORE & BIZ SERVICE (PAYTM), Highly scalable microservices that perform core operations of payments written entirely in Java with Spring Boot, use gRPC calls for inter-service communications and MySQL for storing orders. Implemented event-driven approach throughout the service using Kafka. Maintains payments on various paymodes like Paytm Postpaid, UPI, NET BANKING. Deployed on K8s.
• REFUND DECISION SERVICE (PAYTM), Highly Scalable microservice written entirely in Java with Spring Boot executes refunds by leveraging the power of rule-engine and decides if the customer's refunds are possible to the selected destination. Used Akka framework and non-blocking functionality of gRPC to handle refunds traffic during various banks downtime during peak business hours.
• FPO SERVICE (PAYTM), Java with Spring Boot microservice responsible for fetching available payment modes for merchants. Leveraging an in-house rule engine, the service intelligently determines whether a merchant can accept payments through specific modes based on configured daily and monthly limits. Additionally, it dynamically assesses bank response times and success rates, recommending the optimal payment method for users. With DynamoDB for maintaining bank success rates and Redis for caching responses, we were handling ~12,000 transactions per second on production with a p99 latency of 65ms.
• MEMBER PROFILE SERVICE (EVIVE), Serverless application that uses AWS API Gateway, Lambda Function and AWS RDS (Postgres) written entirely in Node.js. Responsible for fetching the users based on search criteria and tags associated.
• TAGGING CONSUMER-SERVICE (EVIVE), Facilitated tag-based user categorization driven by rules to support a scalable architecture by introducing folksonomy in Evive roadmap. Achieved a seamless transaction to revamped architecture by keeping the old logic alive. Implemented message broker (Kafka) to send the payload to the microservice.

• Spring Boot
• Spring Cloud
• Spring Data
• Spring MVC
• Spring Security
• JPA
• Junit
• Mockito
• Mocha
• Chai

• Microservices
• Maven
• Gradle
• Test Driven Development
• Agile Scrum
• Features breakdown
• Plan Sprints for continuous Delivery
• Monitoring via Prometheus & Grafana