Engineered AI-powered personalization systems using Python, Apache Kafka, and Metaflow to process 1.8M+ daily user interactions and generate real-time homepage recommendations with sub-250ms latency. Developed recommendation workflows integrating ranking signals, feature pipelines, and automated decision systems, improving content discovery by 14% and reducing manual feature engineering effort by 20%. Built backend services using FastAPI, Spring Boot, and GraphQL/REST to deliver personalized experiences while reducing API payload size by 13% and improving homepage load performance. Deployed cloud-native services on AWS (EKS, EC2, S3) achieving 99.9% availability, reducing deployment time by 30%, and improving monitoring/operational visibility. Built distributed streaming systems using Apache Flink, Cassandra, Redis, and Kafka supporting 12K+ QPS with 99.9% pipeline availability. Automated deployments with Docker, Kubernetes, and Spinnaker and used OpenTelemetry/Prometheus to reduce MTT

Software Engineer at CRED India

January 1, 2020 - December 31, 2023

Engineered a distributed microservices-based payment processing platform using Java, Spring Boot/Spring Cloud, and REST APIs serving 8M+ monthly active users with 99% uptime during peak billing. Built event-driven transaction pipelines with Apache Kafka, Avro schema registry, and PostgreSQL processing 12M daily events; used partitioning and consumer-group tuning to keep end-to-end lag under 2 seconds. Reduced p99 API latency from 850ms to 180ms via Redis caching, Hibernate query tuning, HikariCP connection pooling, and database indexing on 20+ hot endpoints validated with JMeter. Containerized services with Docker/Kubernetes/Helm and automated CI/CD using Jenkins, reducing deploy time 65% and moving to daily releases with zero-downtime rollouts. Implemented observability using Prometheus, Grafana, OpenTelemetry, and structured logging, cutting MTTR by 45%. Improved reliability by driving unit/integration testing with JUnit/Mockito and rigorous code reviews, raising coverage from 54% to