feat: Docker support

[volod-vana]

Dockerized app for ease of use, this setup includes a two-stage build process. Initially, it crafts the UI with a sleek Node.js image and Yarn, finalizing UI compilation efficiently. Subsequently, it transitions to a Python environment for app execution, integrating OpenGL support and leveraging Poetry for streamlined Python dependency management.

The build also smartly incorporates the pre-built UI, makes port 8181 available, and gears up for application kickoff with a start.sh script. This approach neatly combines Node.js and Python within a single container, optimizing both ease of deployment and operational simplicity.

Image build

Running in a docker

Additionally, the README has been updated with comprehensive instructions on how to build and run the Docker container, ensuring users have all the necessary information at their fingertips.

docker build -t selfie .
docker run -p 8181:8181 \
  --name selfie \
  -v $(pwd)/data:/selfie/data \
  -v $(pwd)/selfie:/selfie/selfie \
  -v $(pwd)/selfie-ui:/selfie/selfie-ui \
  selfie:latest

Possible Next Steps

• Optimization of Build Times
• Security Enhancements
• Continuous Integration (CI) Integration

[tnunamak]

It would be great if we could utilize any available special hardware (e.g. GPUs).

That said, Docker setups may work best with LLMs that are provided by an external API. Maybe we should document how to use Docker with something like an an Ollama server.