SelAI

SelAI project is a chatbot specialized on programming related subjects.

Before running the project:

Required platform accounts

You need to have these platform account and get authentication tokens:

• Hugging Face - To run DeepSeek model

• ngrok - To run models and access them.

• (Optional) Gemini API - To be able to run Gemini model whenever DeepSeek model is not available.

Required Model inferences

You need to run fine tuned models in a platform like Google Collab or your own server, and set the environment key with a domain:

• NGROK_URL - domain to inference models.

You need to create .env file and copy .env.example contents, set these variables in .env:

• HUGGINGFACE_API_KEY
• NGROK_URL
• NGROK_AUTH_TOKEN
• GEMINI_API_KEY (optional)

Run the application:

You need to have docker and docker compose installed on your machine.

Go into project directory: SelAI run the following command:

docker compose up

The application starts running in: http://localhost:8002 or http://127.0.0.1:8002

Each LLM request is being logged using phoenix. To reach phoenix logs go to: http://localhost:8004 or http://127.0.0.1:8004

(Port can be modified by PHOENIX_PORT environment variable.)