Backend Engineer
Distributed systems, real-time platforms and AI infrastructure.
Python • FastAPI • Kafka • Redis • Go • Kubernetes • Azure

I build backend systems that have to stay fast and correct under load: real-time voice AI, event pipelines, and the platforms AI agents run on. I'm happiest in the hard middle of a system, where concurrency, failover and latency budgets decide whether it holds up in production.

• Real-time voice AI platform: 500+ concurrent calls under 800ms, with distributed scheduling peaking at 500,000 calls/day
• Kafka notification platform: 10,000+ messages/min at 99.9% reliability, with multi-vendor failover
• Multi-tenant AI agent platform on LangGraph: a workflow engine that compiles visual graphs into state machines, and a coding agent that opens verified GitHub PRs
• Search and API performance: OpenSearch over 30M+ records (60% faster search), and 35% faster responses across 100+ FastAPI endpoints

• moderato: async rate limiting for FastAPI, on PyPI. Fixed window, token bucket and sliding window as atomic Lua scripts on Redis server time. One round trip per check, ~8.4k checks/s, 240+ tests and published benchmarks.

• Correctness first under concurrency: atomic operations and one source of time
• Plan for the dependency that fails: retries, failover and backpressure built in from day one
• Measure before optimizing, and publish benchmarks others can reproduce
• Keep it simple until load proves otherwise

LinkedIn · arjunaravind748@gmail.com