AI/ML / 2025

Duplex WebSocket Voice AI Agent

Duplex WebSocket voice AI agent handling client acquisition using Whisper STT, Orpheus TTS, and LangGraph orchestration.

My Role AI Lead & Backend Engineer
Timeline 4 months
Team Size 1 member
Platform Automation

Overview

Developed an autonomous inbound lead acquisition system utilizing a duplex WebSocket voice AI agent. The agent handles natural voice conversations with low latency.

Built a hybrid RAG pipeline featuring Pinecone dense-sparse search, Cohere reranking, and HyDE retries, coupled with a BANT lead qualification process that outputs structured JSON.

The Challenge

Client acquisition teams struggled to respond to inbound leads instantly, resulting in lost opportunities due to delayed follow-ups.

The Solution

Created an autonomous voice agent capable of answering inbound calls instantly, qualifying leads, and booking meetings directly in calendars.

The Outcome

Automated the entire initial sales qualification funnel, eliminating human scheduling delays and capturing clean, structured BANT data for CRM tools.

Core Features

Duplex WebSocket Voice Streaming for Real-Time Conversation
Low-Latency Speech-to-Text (Whisper STT) and Text-to-Speech (Orpheus TTS)
Stateful Conversational Orchestration with LangGraph
BANT (Budget, Authority, Need, Timeline) Lead Qualification Framework
Structured JSON Outputs for CRM & Database Storage
Automated Scheduling Integration via Cal.com API
Hybrid RAG Pipeline with Cohere Reranking & HyDE Retries
Hexagonal Architecture for Modular Integration