DevOps-GPT

Use a brilliant assistant as your friend in your DevOps Journey.

What does this project offer?

This project helps you to balance your daily work as a DevOps engineer, from simple bug fixes to project template generation.
you don't need to search on Google for some routine jobs and it helps you with a robust prompt to simplify your career.

How it works?

This is a wrapper between DevOps engineers and large language models like GPT-4-mini. The main strength of it is prompt engineering.

We have developed some (Extendable) prompt-driven solutions with a simplified interface for the DevOps process cost reduction.

Docker services

We run our services using docker-compose (the main Fast-API-based backend and MongoDB), so we package all dependencies and don't use any external service.
The second one is Mongodb. we need to save the prompt and GPT-4_mini response in the Mongo document for future use. (for example, fine-tuning the other model with our data)

Media directory

The third one is in the media directory. for now, We save a project called (MyTerraform) which is a Terraform template generated based on user requirements. so users can download it and use it. surely we can save any static data in the /media.

This is an example of a template generated in the /media

and the input is something like that

directory_generators

This directory becomes updated when we trigger the template generation API. finally, It generates the MyTerraform directory based on user input which is a template.

ToolChain

1. Python
2. Docker
3. GPT-4o-mini
4. FastAPI
5. Helm

Pre-requisites

1. Python
2. Docker

How to use?

execute sh run.sh in your terminal

Warning

Do Not change the GPT model! Prompts have been developed for the GPT-4-mini model and they can't be integrated with other GPT models. it can cause horrible incompatibility.

Run it local on kubernetes

If you want to run and use this chatbot app within your Kubernetes cluster, you can easily install it using the Helm chart provided in this repository

helm install [RELEASE_NAME] helm/ -f helm/values.yaml

Test

Run tests:

cd app
pytest tests/

Contributing

please read the Contribution guide

Maintenance

** Abolfazl Andalib - abolfazlandalib@gmail.com **
** Mohammad Madanipour - m.madanipourr@gmail.com **