At Gaido AI, I designed and deployed a multi-agent architecture from scratch, with specialized sub-agents and an intelligent supervisor for query handling, policy recommendations, and policy gap analysis.
I built an end-to-end agentic claims pipeline for Health, Motor, and Life. It covered document parsing, OCR, investigation, cited reports, and escalation letters, and analyzed 300+ real claims to contest unfair rejections and deductions.
I built production-grade, real-time multilingual voice AI agents on LiveKit for lead qualification, appointment booking, and policy and claim queries over inbound and outbound calls. I also automated follow-ups across WhatsApp, email, and AI calls, handling around 9,000 messages across 800+ conversations.
I built an evaluation framework using simulated conversations and automated analysis of real chat, voice, and WhatsApp conversations to improve agent response accuracy. At Gaido AI, I also built a daily lead-generation pipeline that generated 200+ leads.
