AI AutomationInternal Product / Remote
AI Voice Receptionist
A real-time voice assistant for incoming calls that answers questions, collects information, and handles conversations in a structured way.
TypeRealtime Voice AI
RoleArchitecture & Development
BackendPython · FastAPI
AccessDemo on request
LIVE / VOICE PIPELINE
TWILIOLIVEKITFASTAPI
My contribution
Ownership from structure through implementation.
The system connects telephony, audio streaming, language models, and backend logic into a modular digital receptionist.
- Connected incoming calls through Twilio and real-time audio sessions with LiveKit.
- Used Deepgram for speech-to-text, OpenAI for conversational logic, and Cartesia for natural speech output.
- Built a FastAPI backend for sessions, conversation context, tools, and external integrations.
- Captured structured information and passed calls or data into downstream workflows.
Technical details
The systems behind the visible result.
- Streaming architecture focused on low latency and smooth conversational turn-taking.
- Modular provider layer for STT, TTS, telephony, and LLM components.
- Tool-based actions for information retrieval and workflow steps during a call.
- Error handling and session management for longer real-time connections.
Technology stack
PythonFastAPILiveKitCartesiaDeepgramOpenAITwilio
Project access
The system is not publicly linked. I can demonstrate it and explain the architecture during a call.
More selected projects
This case study is only one sample. My broader portfolio includes more than 80 websites and several custom automation systems.
Remote · Full-time or part-time
Could my experience be a fit for your team?
I am open to a new remote position on a B2B basis and can share my full CV and further relevant project examples.