Differentially private python web applications
Keeping it short and sweet, we will be focusing on these two topics:
1. How to deploy python web-apps in matter of hours, i.e. create fastly your minimum viable product MVP
2. Whats differential privacy, how does it work, how can we use it off the shelf and incroporate it in our machine learning solution?
We will be moving away from the jupyter-like development enviroment and start serving applications to the consumer.
- Titanic challenge, the famous kaggle challenge will be served and start beeing accessible as a differentially private machine learning solution
- Road trip app, highly versatile app where we try to predict some event (in my case wether the road trip will take place or not) given the weather forecast. But you can use my code from github and modify it to your organisations variant.
- Corona webapp. Given the new situation with this pandemic, we showcased how can you create a product (mortality application) very quickly, and adjust and potentially help people.
For more detailed content and in dept information check out Manuel Amunategui and the ViralML show.