TL;DR the web app is live! Not only can you tap into a library of topping ideas and dough recipes, but you can also unlock the full potential of machine learning to run pizza diagnostics.

Steve Jobs Introduces Pizza Science

I love pizza, especially Neapolitan pizzas! So one day, I got a wood fired pizza oven and started to make my owns. Making pizza is super cheap and easy, but if you want to make good pizza, it then becomes science and it all starts with the dough.

The very first time, I got all-purpose flour and followed a top rated recipe I found on Google. The result was... ok. I had to iterate so it would actually look and taste like Neapolitan pizzas next time.

It took me 6 iterations before reaching the level I was aiming for. I started a Google Sheets to write down my discoveries and document my experiments. I called it "Pizza Science".

This is me running a pizza experiment

Pizza Science

Later on, every time I wanted to make pizza, I would go back to my Pizza Science Google Sheets to make sure everything I was doing was right and to compute the perfect proportions depending on the number of guests I had. It worked perfectly fine, but the interface wasn't really designed for cooking at the same... Also, my keyboard hates flour!

So I decided to make an iOS app with the perfect interface, leveraging my Google Sheets data.

Prototyping

https://www.figma.com/embed?embed_host=notion&url=https%3A%2F%2Fwww.figma.com%2Fproto%2FsLZMjq9LofEllGXSCJlsSk%2FPizza-science%3Fnode-id%3D21%253A115%26scaling%3Dscale-down

Design constraints

• A big-finger interface: because when cooking hands get dirty and taps become inaccurate
• No scrolling: we limit the information we display to what's needed when cooking
• Straightforward: complete the lists and the app walks you to the next step

Technical constraints

• No backend, for a faster development
• Client-side
• iOS only

Database - Airtable API

No backend doesn't mean no database. For this project, I used Airtable to store the recipes and attached a link to the pictures on Imgur. I then used the API to get the data in the app.

https://airtable.com/shrsLpVgE9KD8NXXF

What I learned

This project was imagined and developed in under 2 weeks. I will probably be the only user as I'm not planning on marketing it and there is no real need, but I learned many things:

• It's the first iOS app I push on the App Store myself so I'll be faster next time
• Minimalism is key to deliver a MVP at full speed, I have 20 more features in mind but kept only the critical ones for a super fast launch
• User interface choices must be radical for consistency
• Also, my pizzas reached a new level! (if you hire me, I'll let you eat them)

Why is there machine learning?

A few weeks after I finished the project, I explored the capabilities of TensorFlow and tried their image recognition models. Inspired by the show Silicon Valley (Hot dog / Not hot dog), I thought it would be fun do this with pizza. It is trained with 1000 images of pizza and 1000 images of other types of food.