How Google enabled me to create an AI powered app in less than 40 hours of dev time.
The company I work for, Hi5, recently sent me on a GCP (Google Cloud Platform) on-boarding course, as one of our values is to be constantly learning by being “curious.” The focus of the full-day course was to introduce Google’s latest technologies in Data Science, Artificial Intelligence and Big Data.
I’d already delved into the world of AI in my own time using Python and TensorFlow, but was not yet familiar with their incredibly robust set of existing APIs and pre-built training models. Though we didn’t go into much detail about these tools during the course, it did spark a further interest in me to make use of their tools (Why waste time creating brand new models to do tasks they already have the models for?).
This is what led me to do a trial run using another cool Google-backed technology, called Flutter, and my newly gained knowledge of their GCP.
These are the steps I took:
1 Created my account on the GCP at https://cloud.google.com/
2 Enabled the tools I wanted to use in my app with GCP, generating access keys, etc.
3 Created a server droplet on Digital Ocean, which I used for authenticating with GCP and running my server code.
4 Wrote my server code in nodeJS and an endpoint to receive images from users’ devices.
5 Created a simple, single-view app in Flutter that allows users to either upload an image from their gallery or snap a photo with their phone camera, and in turn receive classifications of their image with confidence percentages for each classification.
6 Back-and-forth coding/debugging/testing until I was happy with the results (for now) 😏
7 And lastly, publishing the app to the Play Store — see it here: Waila (What Am I Looking At).
I managed to get all of this done in my free time, and it took less than 40 hours of development time to get it to where it is at the time of writing this.
I could have spent more time adding fluffy UI/UX things and some more functionality to the app, but this side-project was intended to be nothing more than a test of the capabilities of the GCP and Flutter, and I’m more than happy with the result.
I have continued to look deeper into their tools and all the lovely ML goodness that Google has so generously made open source, and can unequivocally state that it is a game changer for the world of Machine Learning. I am thrilled to be able to be a part of it.
If you’d like to check out the app you can find it here:
We’ll soon be incorporating AI into Hi5, too, so be sure to keep an eye on it!
Clap if you found this interesting 😉
