--- a
+++ b/README.md
@@ -0,0 +1,22 @@
+# deploy-gradio-ml-on-gcp
+A demo ML application that provides first aid information to people diagnosed with internal bleeding.A simple ML application built with HuggingFace and Gradio Deployed on GCP
+
+## Live Demo
+> https://ml-gradio-demo.ew.r.appspot.com/
+
+## KaggleX Mentee - My Experience
+> [LinkedIn Post - My experience as a KaggleX Mentee](https://www.linkedin.com/posts/emmanuel-katchy_kaggle-kagglex-bipoc-activity-7126311581159727105-QzS_?utm_source=share&utm_medium=member_desktop)
+
+## KaggleX Program Website
+> https://www.kaggle.com/kagglex
+## Helpful Tutorials
+ - https://medium.com/@mehmetcanfarsak/ci-cd-using-github-actions-to-deploy-on-gcp-app-engine-in-3-simple-steps-399fa9764cf2
+
+### Notes/Todos:
+  - You need to disable Gradio flagging in GCloud because the environment is read-only
+  - Allow users to upload scan images, and perform live inference
+  - To be able to use this template/project, create a blank `env.yaml` locally. Environment variables can then be defined as needed.
+    An example:
+  ```yaml
+  HUGGING_FACE_TOKEN=<YOUR_API_KEY>
+  ```