AIden

Copilot, akin to Jarvis from Iron Man. A project inspired long before the GPT days. Uses OpenTelemetry for vendor-agnostic telemetry, forwarding metrics, logs, traces and spans to a collector.

Software Versions

• Python3: 3.10.12
• Venv: python3.10-venv

Test

To run the pytest suite, navigate to the root directory of the project and run

pytest

Linting

The linter used here is pylint, to run just execute

pylint <path-to-file.py>

Usage

Docker Compose

First Time Setup

Install the NVIDIA drivers for your card before running the commands below:

chmod +x ./bootstrap/nvidia-cuda-toolkit.sh
bash ./bootstrap/nvidia-cuda-toolkit.sh
chmod +x ./bootstrap/nvidia-container-runtime.sh
bash ./bootstrap/nvidia-container-runtime.sh
sudo systemctl daemon-reload
sudo systemctl restart docker

Subsequently

docker-compose up -d

Integrating your implementation with monitoring

The OTel collector is primed to accept metrics, logs and traces on http://localhost:4317. At the minute, it does so over gRPC. You can enable HTTP if you need it. Any code should use that as a target to send to.

Extras

The deployment includes Homepage, an application dashboard. To make use of it:

• Navigate to

http://localhost:8092

The dashboard performs service auto-discovery so if you add a service to the stack, and label it the same way the other services are, it will show up in the dashboard automagically.