Built a production-ready real-time voice agent integrating Twilio and OpenAI for conversational AI automation. Developed a cost-effective speech-to-speech solution using Sarvam AI, ElevenLabs, and Deepgram APIs, reducing infrastructure costs by 60%. Implemented dynamic tool invocation with RAG pipeline for document-based context and optimized performance to reduce response latency by 50%.