This is the repository for the workshop "AI-assisted Newsletter Dashboard" that was held during the datalift summit. It is self-containing, such that you can reproduce all steps from the workshop :)
To do so, please also look into the file slides.pdf. If you have any further questions, please don't hesitate to contact:
- moritz.feuerpfeil@kern.ai, main contributor to this workshop
- johannes.hoetter@kern.ai, co-founder of Kern AI
In this repository you will find all the necessary files to follow along. This repo is organzied as the following:
- The folder structure represents our workshop outline, starting with 01_MailExport
- The directories where we code are organized in three sub-categories
- scratch: necessary files but only filled with comments if you want to code along the whole journey
- guided: necessary files with some code filled in for a more relaxed follow along
- finished: the solution we provide for the next step
- We provide all the necessary data for the following steps in the finished sub-directories
- clone this repository or download it as a zip
- (if you don't have anaconda installed) install anaconda
- (recommended) create a new conda environment
$ conda create --name datalift python=3.8 $ conda activate datalift - install the requirements
$ pip install -r requirements.txt
We strongly recommend using VSCode as your IDE.
Necessary Extensions:
- Jupyter
- Python
Suggested Extensions:
- Rainbow CSV
- Material Icon Theme
To run the fastAPI backend and streamlit frontend, please follow these two steps:
- go into 06_FastAPI/finished and type
or (if the previous does not work)
$ uvicorn main:app --reload$ python -m uvicorn main:app --reload - go into 05_Streamlit/finished and type
$ streamlit run app.py
You're now up and running! Make sure to use streamlit in "wide mode", which you can change in the settings that you can access on the top right of your UI.
The app developed in the 3 hours workshop is not perfect yet. Take the following list as an example for things on how you can improve the app further on your own:
| Topic | Time |
|---|---|
| 📧 collecting data with oauth2 in GMail | 15 min |
| 🔧 processing data (e.g. HTML and text parsing) | 30 min |
| 🔖 building training data for topic modelling | 15 min |
| 🧠 applying transfer learning with transformers and recommenders | 30 min |
| ☕ break ☕ | 10 min |
| 👑 building a streamlit UI | 40 min |
| 👾 building a minimal backend via FastAPI | 40 min |