
AI Restaurant Ordering System (Work in Progress)
I am developing a full-stack AI-powered restaurant ordering platform that integrates telephony, speech recognition, and natural language processing to automate phone orders. The system is built on VitalPBX/Asterisk, using SIP for call handling and Faster-Whisper for real-time speech-to-text transcription. Customer voice input is processed by an AI assistant that interprets the order, manages menu logic, and generates conversational responses through text-to-speech.
On the backend, I have implemented a Python service (Flask/FastAPI) that connects directly with the PBX through ARI (Asterisk REST Interface). This service currently handles authentication, call answering, recording, transcription, and order flow management while securely storing menu data and transcripts.
At this stage, the system is fully functional for a single restaurant, with successful call handling, AI-driven transcription, and dynamic menu-based ordering already achieved. The next milestones focus on expanding toward a multi-restaurant SaaS platform, with features such as individualized menus, analytics dashboards, and auto-scaling for peak hours.
This project is still a work in progress, but I have already built and tested the core components that prove its viability. It demonstrates my ability to connect low-level telephony systems with modern AI models, build production-ready APIs, and design scalable solutions from the ground up.