Apollo Core

The Apollo Core turns any home into a smart home with AI voice and lighting automations. Everything runs locally with no cloud or WiFi dependence. The Core is private, fast and personalized to you.

Comment 2

Inspiration

Our team wanted to build a smart home system that was private, reliable, and fully local. We've all used assistants like Siri and Alexa, and we saw the same issues. Slow responses, cloud dependence, and devices that did not work together. We wanted a system that lived inside the home, protected user data by default, and responded instantly.

This project started with a shared goal. Build a local smart home core that a household could trust. The Apollo Core is our first major step toward that vision.

What It Does

The Apollo Core runs a local voice assistant, lighting control, and a secure API on a Jetson Nano. All processing happens inside the home. The system lets users control lights, run automations, and interact with the assistant through a React Native app or a web dashboard.

How We Built It

Our goal with this stack was to build an AI Assistant device that runs offline, stays private, and still feels modern and responsive.

  • We compiled an older version of llama.cpp to handle the Jetson Nano's hardware limits.
  • We used Whisper.cpp for our Speech To Text and Piper for our Text To Speech to generate fast and clear local speech output.
  • We deployed Home Assistant in a Docker container and integrated four GE Cync smart bulb through its local API on the Jetson Nano.
  • We built a custom REST bridge that connects the app to Home Assistant without exposing secrets.
  • We created both a mobile app and a web dashboard using React Native and TypeScript.
  • We set up local networking so the system works completely inside the home with no external dependencies.

Challenges

  • The Jetson Nano's limited compute required careful tuning of llama.cpp and Piper.
  • Home Assistant, Docker, and older system libraries needed a lot of compatibility work.
  • Browsers blocked direct local requests, so we had to solve CORS and mixed content issues.
  • Routing traffic safely through HTTPS required reverse proxy configuration and DNS fixes.
  • Every component needed to communicate on the same LAN without exposing secure tokens.

These challenges pushed us to understand every part of the system from hardware to network routing to software.

What We Learned

  • Running AI models and TTS engines on constrained edge hardware
  • Integrating Home Assistant with custom APIs
  • Managing local only authentication and secure token handling
  • Building consistent mobile and web interfaces for local devices
  • Designing a scalable home automation architecture that preserves privacy

What Is Next

We plan to expand Apollo Core into a complete local smart home platform. Our next steps include:

  • Local wake word detection
  • Multi device networking inside the home
  • Advanced automations and routines
  • A unified Apollo smart device ecosystem
  • A faster voice pipeline with better audio processing

The Apollo Core is the foundation for a privacy first smart home system that households can truly own and trust.

Built With

Share this project:

Updates