Sep 2024 – Present · 1 yr 8 mos · Kolkata, West Bengal, India · Hybrid

• Architected and implemented a full end-to-end integration between Vapi’s
• Voice AI Agent and Dentally CRM using FastAPI, Webhooks, and MCP.
• Designed REST APIs that allow Vapi to directly trigger backend workflows
• such as fetching appointments, booking slots, updating patient data, and
• validating dentist schedules in real time.
• Implemented secure HMAC webhook verification, raw-body logging, and
• asynchronous processing to ensure reliable communication between Vapi
• and backend APIs.
• Leveraged Dentally API (patients, appointments, practitioners) to automate
• the entire booking flow without manual intervention.
• Built a flexible conversation → action mapping engine, enabling the Vapi
• agent to execute custom backend operations using your internal logic.
• Added support for real-time event handling (incoming calls → API actions),
• improving booking accuracy and reducing human intervention by 100%.
• Containerized and deployed the API with proper environment handling,
• async logic, and error-resistant integrations. Deployed it to AWS EC2.
• Fixed and improved EC2 deployment for the production Vapi agent backend:
• Set up Nginx reverse proxy + Uvicorn workers
• Enabled HTTPS and static IP exposure
• Ensured 24/7 uptime for Vapi webhook URLs
• Deployed environment variables + secure credential handling
• Implemented continuous logging and observability (request IDs, session IDs, appointment trace IDs) to track real-time booking flows across Vapi → FastAPI → Dentally.
• Created a debug-mode audit system that records all appointment lifecycles—slot checks, practitioner availability, Dentally payloads, and webhook intents—to eliminate silent failures.
• Improved overall accuracy of the automated appointment booking system, reducing appointment errors by >95% and ensuring consistent bookings for correct patients.