Anish Agrawal

I’m a backend engineer focused on building high-scale messaging and payments systems for customer communication. I’ve worked on reliability and scalability challenges that handle massive event volumes—up to 500M+ events/day—while keeping failure behavior predictable.

At Gupshup, I designed a distributed retry and timeout orchestration system (processing ~25M events/day) to support cross-channel failover and bounded backoff retries. This work reduced the system wide failure rate from 8% to 5%, and I also built a failover-aware JDBC wrapper for Aurora with ProxySQL-backed MySQL clusters, cutting recovery time from 15 minutes to 30 seconds.

I re-architected the Redis messaging state layer to sustain peak throughput of 150K+ events/sec by using batching and backpressure, consolidating Sentinel clusters into a 3-master/3-replica Redis Cluster and reducing infrastructure footprint by ~66%. I led staged SMPP infrastructure migration under live traffic, using a zero-loss cutover strategy and data-driven rollout sequencing to avoid service disruption.

Before that, I rebuilt message ingestion from Java 8/Servlets to Java 17/Spring Boot on Kubernetes, enabling containerized deployment with behavior parity through strong test coverage. I also delivered async WhatsApp payment workflows (500K+ interactions/day), integrating webhook events and migrating sync APIs to async ingestion, while building operational dashboards with Go and Retool to improve payout visibility for production teams.