Poornachandra M Nayak

I’m a Senior Backend Engineer focused on building production-grade voice AI platforms with scalable backend services. I architect async Python/FastAPI microservices that support thousands of daily voice interactions with high reliability.

In my current role at Orbitel Labs Pvt Ltd, I’ve powered a voice AI system serving 5,000+ daily calls across ~60 bots with 99%+ reliability under high-concurrency load. I re-engineered a RAG pipeline (ingestion → normalization → retrieval) to cut p50 latency 40% (150ms → 90ms), improving voice-bot response quality at scale.

I also designed an LLM tool-calling framework with strict JSON schema validation to enable complex multi-step actions mid-conversation—without hallucinated function calls. To handle real-time demands, I scaled the retrieval backend to sustain 80–100 concurrent AI queries during live voice sessions using async workers and optimized vector query patterns.

Earlier at ICICI Lombard, I delivered high-throughput Python/Flask APIs and AWS-integrated services, shipped a robocalling platform to production, and increased conversion rates by 30%. Before that at Unbox Technologies, I built Go services for real-time CDR synchronization, redesigned for multi-tenant architecture with Docker and Lua routing to cut infrastructure costs 35%, and improved call-routing conversion rates by 20–25%.