Notes from an architect who still reads the postmortems.
I design event-driven data platforms for regulated financial services.
Systems that have to be correct, auditable, and fast at the same time. That means Kafka, Confluent, and the operational discipline to run them in production.
Lead Engineer · Rocket India · Chennai
topic: customer.events · 4 partitionslag: 0
500+ tables unified50+ source systems1M+ API calls at peak hour5,000+ topics governed
Writing
May 9, 2026 · 14 min readData platforms at scale
Confluent Cloud vs self-hosted Apache Kafka: real pricing, engineering cost, and a decision framework for choosing managed vs DIY Kafka in 2026.Kafka to Snowflake: real-time ingestion patterns at scaleMay 7, 2026 · 13 min readData platforms at scale
Snowpipe, Snowpipe Streaming or the native connector: getting Kafka data into Snowflake at the latency and cost you actually need.Five Kafka patterns for event-driven microservicesMay 4, 2026 · 16 min readEvent-driven architecture
Five event-driven patterns with Kafka: when each one fits, how to implement it, and the failure modes that cost teams months in production.All writingSelected work
Rocket India, US mortgage servicing, 2021-2025Unifying 500+ tables from 50+ source systems into one real-time customer viewA streaming customer data platform built without a freeze window on any source system, serving 1M+ API calls at peak hour in real time, with a Snowflake gold layer that cut a common downstream query from about three seconds to roughly 500 milliseconds.
All case studiesKafkaConfluentDebezium CDCksqlDB
Rocket India, Kafka platform, 2019-2026Making Kafka self-service for 200+ teams without letting any of them near production dataOne control plane where naming, retention, ownership and access are enforced at request time rather than agreed in a document, still in production and governing 5,000+ topics.KafkaConfluentKubernetesFlink
Rocket India, Confluent Cloud migration, 2025Migrating a production Kafka estate to Confluent Private Service ConnectEvery producer and consumer moved to private networking incrementally, team by team, with zero data loss and no re-keying of existing topics, alongside a new real-time data exchange platform for secure, cross-cluster sharing.Confluent CloudPSCCluster LinkingTerraform
Stack
Streaming
Kafka · Confluent Cloud · ksqlDB · Kafka Streams · Debezium · Schema Registry
Platform
Kubernetes · Terraform · Helm · GCP · AWS · Azure · Cloud Run · GitHub Actions
Services
Java · Spring Boot · GraphQL · Python · Node · PostgreSQL · Oracle
AI systems
Agentic orchestration · retrieval pipelines · evaluation harnesses · LLM cost and latency budgeting
If you’re building something that has to be correct at scale, I’m happy to talk it through.
Start a conversation