Learn By Doing Volume 28 -- Turn old pictures into modern representations using Machine Learning

Learn By Doing Newsletter

A free weekly curated cloud, blockchain, and coding newsletter delivered to your inbox every week.

Blockchain Newsletter

Subscribe to get the best cloud technology, blockchain news, and coding articles in your inbox every week

We will never send you spam and it's easy peezy to unsubscribe at anytime.

☁️ 📖 Learn By Doing Volume #28 💻🔨

This week was quite the week for the Learn By Doing newsletter, it went viral on Hacker News for a little bit. It's always cool to see a project you are working hard on get some recognition and attention.

With that, welcome to Volume 28 and welcome to all the new subscribers out there! This week we have some cool stuff in the Cloud space, like how to manage a Kubernetes cluster that uses spot instances. We also have yet another way to get PHP running inside of a Lambda function. There is also a nice peek behind the curtain at how the cost based optimizer is implemented in Cockroach DB.

This week, the Cool Find might just be my favorite of all time. DeOldify uses machine learning to turn old photographs into modern day color works of art. It's simply incredible, check it out.

☁️ Cloud

Amazon API Gateway Adds Support for AWS WAFWeb Application Firewall (WAF) is a very underrated service in my opinion. It provides some slick integrations at the edge of our network and allows us to up our security game. Now all of that can be combined with APIs that are running in API Gateway.

Embracing failures and cutting infrastructure costs: Spot instances in KubernetesRunning K8s clusters can certainly get expensive, so folks are devising ways to bring the cost down. Here is a nice introduction into some fundamental tradeoffs that are made when thinking about spot instances and how they can be incorporated into your K8s architecture.

Setting up a Serverless PHP Website with AWS Lambda and BrefFor every language that is not supported by AWS Lambda there is sure to be someone trying to figure out a way to get it running in the serverless world. This post dives into how you can write PHP, package it up as a binary, and then run it via Node in a Lambda function.

🔨 Tools

eventstore - Functional database with Complex Event Processing in JavaScriptEventStore is a database that enables you to write thousands of events a second and read them almost as fast. This is an interesting idea that extends off of the CQRS pattern which allows us to signal to other services via emitted events. I'm not sure I would have a strong need to run this type of database myself as I can get something similar using AWS SNS topics, but it is cool nonetheless.

GraphQL cheatsheetGraphQL is very convenient but sometimes the syntax can be tricky to remember. Luckily, there is this handy cheat sheet to keep open in a tab in order to check yourself. Should save a few comments on a PR or two.

spectre - Another lightweight, responsive and modern CSS frameworkI'm still pretty excited about all of the things I am learning regarding Tailwind CSS, but I just came across Spectre as well. This looks like another, more lightweight, version of a non utility first framework. There is some neat things in here that I haven't seen other frameworks do that make this worth a look.

Code && Languages

Atari - Reinforcement Learning in depth 🤖Remember Atari games and how addicting such simple pixel based games could be? Well now read up on how you can leverage some machine learning to automatically play the game for you.

An in-depth look at 100% Zero Downtime deployments with TerraformPerforming zero downtime deployments using Terraform is more complicated than you might initially think. Here Checkly explains their process for deploying the latest workers without taking things offline. Keep in mind there is a few primitives built into Terraform that can help you do this, but they will only take you so far.

How We Built a Cost-Based SQL OptimizerThis is an interesting view into the world of Cockroach DB and the SQL optimizer they built. They evolved their heuristic based optimizer into a cost based optimizer that takes the resources and time of each execution plan into account. There is a lot to take in here but there it is also a very interesting read, especially when you get a peek under the hood.

😎 Cool find of the week

DeOldify- A Deep Learning based project for colorizing and restoring old imagesThis is not an introduction to machine learning resource, but rather a very complicated, practical, and just downright cool use of it. The name kind of summarizes what it does, but give the README a read and you are going to be impressed by the work put into this.

© 2019 Kyle Galbraith. All Rights Reserved.