When I first started this blog, I found it quite tricky to have Jekyll up and running properly.
I followed the guide in Github and the guide in jekyll’s own website, but kept having problems.
These problems were related to dependencies, the gems ruby uses, and installing them wasn’t an easy task.
When you solve a problem, or get something working, do you ever feel “dirty”? Like the solution you
found wasn’t ideal, and your environment is polluted? Or that strange feeling you get, when something works but you don’t exactly know why?
I absolutely hate that feeling… And I felt that way, until I found Docker.
If you have checked my LinkedIn, you know I am a Data Engineer at Cytora.
Part of my job is to work in our product, the API that serves our AI models to our clients.
Part of a good product is to be able to provide the users with a good experience while they use it,
or develop tools that interact with it. To do so, you need to be able to tell them exaclty what is going wrong.
A simple 400 Bad Request won’t work, you need to be able produce custom error messages and codes.
However, what should be an easy task, isn’t.
I have a Dell XPS 13 9360, and I have just started using i3wm.
Installing Ubuntu and i3 was easy enough following the tutorial on i3wm’s website.
One of the hard parts was getting the brightness keys to work.
The first post established all my project objectives, so this post serves to give a bit more structure to the idea.
For this project, it is my personal preference to keep everything in the same repository, and well separate.
This is my first real project. I have a plan and I intend to stick to it. This project is called “Fake Foods” and is inspired by Artificial Intelligence, Natural Language Processing and Generation, and the “fake news” concept that has taken over the world in 2017, especially after Trumps election.
What is Fake Foods?
The plan for this project is to create recipes using AI. The point is not to discover new, amazing recipes, only to generate recipes that can be read and make sense. I will call these “Fake Recipes”.
These recipes will be published in a website, and I want to see if it gets any traction at all and if people comment on them. The fake foods website will need to look real, and for that magic to happen, I will also add AI generated comments to each of the recipes.
At the end of the project, I want to have developed my skills in Data Mining, Data Cleaning, some Data Science and Machine Learning, and web development.
As deliverables, I intend to make available a model and API delivered (containerized with Docker) that people can use, supplying a title, or ingredients, and the API will generate a recipe. Is this doable? I don’t know, but I will try.
What is the plan?
For this to happen, I will take the following steps:
- Scrape recipe URLs off several food websites;
- Scrape the ingredients and methods each recipe has;
- Explore the obtained data using best practice with Data Science Cookie Cutter;
- Develop a Flask API to deliver the model created;
- Containarize the app and release it into the world;
- Develop the website to post all the fake food.
Slowly but surely I want to make this happen.
If you want to contribute, let me know!