krishna kumar

I’m a Principal Engineer with 12 years of experience in AI/LLM Platforms & Distributed Systems, specializing in Java, TypeScript/Node.js, Python, and Go. I lead teams to turn complex platform problems into reliable, measurable capabilities—especially where performance, security, and developer usability matter.

At Zuora, I defined the AI platform strategy on AWS Bedrock, improving conversational quoting while implementing safety guardrails that reduced critical workflow errors by up to 40%. I built an evaluation harness that lifted LLM tool-calling accuracy from 61% to 97.9% and architected an MCP-based platform to expose CPQ operations safely as LLM-callable tools across 15+ agents. I also shipped a multi-agent orchestration platform with tenant-filtered entitlement checks, reducing agent onboarding time by 30% and effort by 15%.

Earlier, I led a team of 11 engineers at Exotel to build a business communications platform adopted by 10,000+ businesses and supporting 300,000+ daily active users. I re-architected critical production systems to reach 99.9% uptime and reduced service delays 80%, cutting AWS costs by $42,000/year. I’m especially motivated by engineering pipelines that improve latency and scale—like collapsing dashboard latency from ~15s to sub-second for 2,000 tenants.