Year
2024
Services
Summary
AI-led care-provider supply and operations at Emoha — chat and voice agents that onboarded 33,000+ carers pan-India, a dual-model Graph RAG pipeline tuned for token economics, and the ops modules managing live cases and ₹8 Cr+ in monthly payouts.
View site
Overview
Role
Emoha's home-care supply ran through vendors — a margin cut on every deployment and no control over quality or availability. Going direct meant onboarding carers at a volume no human team could affordably handle: high applicant counts, many languages, most applying by phone. The fix had to be AI-first.
I engineered the onboarding funnel: conversational chat and voice agents that took carers from application through KYC, at a scale that passed 33,000 onboarded carers across India. Under the hood, a Graph RAG retrieval layer with dual-model routing — a smaller model for routine turns, a stronger one only where reasoning demanded it — because token cost per conversation is the unit economics of a voice funnel at this volume.
Around the agents I built the operations backbone: the Home Care module giving ops a real view of active cases, attendance input and tracking so service delivery is evidenced rather than asserted, and an expense module handling ₹8 Cr+ in monthly vendor payouts. Together the system lifted margin by 7% and turned deployment oversight from an intention into a record.



Tech Stack
Next project
Software Development Engineer, Full Stack Developer, Web3 Engineer, Next.js Enthusiast, Freelancer,
Software Development Engineer, Full Stack Developer, Web3 Engineer, Next.js Enthusiast, Freelancer,