In this post, I share the details about the infrastructure I use to run the following websites:
- https://www.katnoria.com: The primary website where I post long form articles
- https://kb.katnoria.com: Mostly contains short posts with code snippets
- Sprinkle AI: Old domain where I host some fun apps
- https://faces.sprinkleai.com: Illustration of potential Facial recognition applications from 2018 post
- https://text.sprinkleai.com: Pre-Transformers era language models
Static Website hosting. It doesn’t get easier than this. You can maintain the website code on GitHub and configure Netlify to dynamically publish the changes on every commit. Netlify does all building and deployment. It also provisions the SSL certificates 😎
I run all my backend code on DO(DigitalOcean). Back in 2016-2017, I explored AWS and Google AppEngine but settled for DO because of its intuitive user interface and fixed fee based model. And, I still see no need to move to any other cloud for hosting simple services.
For serverless, I would use AWS though.
I have been using GitHub for last 4+ years and it is now my default cloud based version control system. Prior to that, I used self hosted GitLab but eventually decided to move away when one of the HDD crashed. Luckily, the NAS was running on RAID 1 so no data loss but something had to be done.
Google Colab Pro
I would argue that it is a must have resource for anyone doing Deep Learning. You get P100 or a TPU depending on what you want to use. It is a hassle free environment to do experimentation with DL models in the cloud (its not the only way but its good enough for my usecase).
TODO: Expand more on this
I have used both Azure and AWS at work but for personal use I just use the block storage to store static content.