Add a CONTRIBUTING.md file. (#62)

* Add a CONTRIBUTING.md file.

* Remove email

CONTRIBUTING.md

@@ -0,0 +1,30 @@
+# Contributing to Sage
+Thank you for considering contributing to Sage! We welcome all kinds of contributions. You don't need to be an AI expert to have meaningful impact. If you have any questions, reach out to us via [Discord](https://discord.gg/zbtZe7GcVU).
+
+If you are looking for an internship or full-time job, becoming a contributor is the best way to surface to the top of our resume stack!
+
+## Ways to contribute
+We welcome your help in multiple directions. Feel free to pick the one that best matches your skills and interest:
+
+- **Improving documentation**. All our documentation is currently in the project [README](README.md). If you find the instructions incomplete, unclear or confusing, let us know via [Discord](https://discord.gg/zbtZe7GcVU), submit an issue, or send a PR with improved verbiage.
+- **Simplifying the setup**. All setup instructions are documented in the [README](README.md). Currently, setting up the environment is more cumbersome than we'd like. For instance, the local setup requires installing Docker and [Marqo](https://www.marqo.ai/), while the remote setup requires you to create accounts with various providers like [Pinecone](https://www.pinecone.io/). We welcome suggestions for streamlining it.
+- **Plumbing tasks**. Under the "issues" tab, you will find feature requests marked with difficulty tags between 1 and 5. As a rule of thumb, difficulty labels 1-2 refer to easy coding tasks that don't require any knowledge of AI. They are meant to improve the engineering pipeline without affecting the quality of code retrieval or chat responses.
+- **AI quality improvements**. For contributors familiar with AI, these are the most fun feature requests, marked with difficulty 3-5: they encourage you to experiment with state-of-the art techniques for code retrieval and text generation. We welcome code contributions but also suggestions for relevant papers we should check out.
+- **Evaluation**. We are currently using our own benchmark for retrieval (more details [here](benchmarks/retrieval/README.md)), but we welcome suggestions for additional datasets, evaluation metrics or evaluation pipelines.
+- **UI/UX**. The chat experience is surfaced via a barebones Gradio app. We want to hear your suggestions for more practical or prettier form factors.
+- **Branding and marketing**. If you have suggestions for better imagery (logos / header image), messaging or social media presence, we're all ears!
+
+### Code contributions
+For code contributions in particular, we suggest the following workflow:
+- Fork the repository
+- Clone the repository locally to your machine
+- Make changes and commit them
+- Push the branch to your local fork
+- Submit a pull request with the described changes.
+- If you are addressing an existing issue or feature request, make sure to reference it under the "Development" section of the pull request.
+
+### Non-code contributions
+- If you simply have suggestions for improvement or marketing materials, let us know via [Discord](https://discord.gg/zbtZe7GcVU).
+
+## Hacktoberfest 2024
+We welcome Hacktoberfest contributions for all the items above. Issues that we consider particularly beginner-friendly are marked with the "hacktoberfest" label.