• Redesigned legacy UGC (Reviews & Ratings) system into a scalable, distributed, high-availability, fault-tolerant, low-latency cloud-native architecture.

• Designed and developed containerized microservices (Java, Spring Boot, GraphQL-Apollo) on Azure/GCP, leveraging patterns like API Gateway, Circuit Breaker, Sidecar, Service Discovery, and DB-per-Service.

• Reduced service latency by 3× while sustaining millions of transactions per minute, using precomputation, sharding, caching, and next-gen infrastructure migration.

• Built high-performance search services (<15ms) with Elasticsearch, including semantic search using vector embeddings, optimized sharding, replication, caching, pre-index sorting, and low-cardinality querying.

• Implemented Model Context Protocol (MCP) for efficient context management in LLM inference pipelines, improving AI model performance and reliability.

• Leveraged generative AI tools to summarize reviews, extract key aspects, sentiment, and brand voice, enabling actionable insights and enhanced content quality.

• Optimized system throughput and efficiency through advanced caching, prefetching, connection pooling, compression, async processing, and Kafka optimization (parallelism, batching, partitioning, Protobuf ).

• Designed and optimized database schemas and data models for Cassandra, Elasticsearch, and Cosmos DB, implementing indexing, sharding, partitioning, replication, and query tuning to ensure high-performance, scalable workloads.

• Built scalable ETL pipelines with Spark, Elasticsearch, and Airflow, handling terabyte-scale data ingestion and transformation.

• Partnered with Data Scientists to design and deploy ML models into product features, including prompt engineering and LLM output optimization.

• Designed and developed scalable, high-availability microservices (Spring Boot, Java 8, REST) on AWS for mission-critical workloads.

• Built Search & Discovery services handling 1M RPM with 20ms 99th percentile latency, including geo-sharded Elasticsearch and inverted indexes.

• Designed and optimized database schemas and data models for MongoDB, MySQL, PostgreSQL, and DynamoDB, with indexing, sharding, partitioning, replication, and query tuning for high-performance workloads.

• Developed orchestrators and event-driven microservices (RabbitMQ, gRPC, ProtoBuf) for complex workflows and asynchronous data flow.

• Built large-scale data pipelines with Storm, Spark, Kafka for ingestion, analytics, and ML.

• Leveraged modern data platforms: Snowflake, Neo4j, Timescale DB for reporting, graph use cases, and real-time analytics.

Optimized JVM performance through GC tuning, streams, lambdas, Completable Future etc...

• Integrated SAML & OAuth for secure identity and access management.

• Designed and developed highly scalable and HA SaaS applications on cloud.

• Used Asynchronous programming techniques, including multi-threaded application design and development.

• Architect stable, efficient, and scalable cloud platforms.

• Develop architectural diagrams to illustrate architectural complexities and interaction.

• Written Lua scripts at Redis Cache server for filtering cache for search and heavy cleanup of cache.

• Used MYSQL MongoDB for multiple use cases and tweaked to improve performance and efficiency.

• Implemented OAuth2 and SAML for secure authentication and authorization across distributed microservices.

• Designed and developed performance oriented multi-threaded connectors for extracting Peta bytes of data and transferring data from client systems to the HA resilient distributed application servers using TCP/IP keepalive connection and transfer tera bytes of data. The systems are built to transfer data using custom Application Layer Network Protocol (over TCP socket) to communicate between servers and agent to enable scalable data transfer. Written Lua scripts at Redis Cache server for filtering cache for search and heavy cleanup of cache.

• Designed and developed highly scalable jobs Spark streaming jobs on HDFS platform integrated with Kafka.