At Gnani.ai, I architected and delivered the vachana.ai backend infrastructure end-to-end, enabling enterprise-grade speech services with 99.9% uptime. I build scalable STT/TTS APIs across REST, SSE, WebSockets, and gRPC, handling 1,000 RPS.
I designed event-driven systems using PostgreSQL and RabbitMQ that process over 10 million daily events for tracking, billing, and authorization with 99.99% transaction accuracy. I also integrated NVIDIA Triton and NVIDIA Riva inference engines for sub-200ms real-time speech processing.
Previously at Sowiz Inc, I built referral and RBAC capabilities, streamed user events with Kafka, and migrated observability from New Relic to OpenTelemetry, reducing monitoring costs by 40%. I work primarily with Go, Python, Docker, cloud platforms, databases, and messaging systems.
