I’m a software engineer currently working in data and machine learning.
This website is gradually becoming my static repository for all written content, in pursuit of ditching more and more social platforms. For specialised work like code, you should still visit my GitHub profile, everything else will eventually find a home here.
Applications that depend on databases and other services make for fiddly local setup and Docker Compose is a common solution, but what about IDE integration? This article shows how you can develop Django applications entirely within a container using VS Code and Dev Containers.
Whenever a shared machine, such as a build server, needs access to your GitHub organisation, we traditionally opted for personal access tokens (PATs) or SSH keys created against a machine user. This works until you consider the security implications of not being able to attribute actions made by that user to any of the humans with access to it.
Imagine you have an internet provider. You probably do. This provider sometimes performs scheduled maintenance, resulting in your internet temporarily disconnecting. Now imagine there's no way of getting notified about future maintenance windows, unless you check their outages page every day. Annoying, but we can do something about that with a bit of code.
I happen to be an aviation nerd and flight tracking services enable aviation nerds to learn a lot about any aircraft flying over. Where it's going, where it's coming from, why it's flying over my house at 6AM after I went to bed only 4 hours ago. Important stuff.
Many (if not most) companies reach a point when data becomes a priority. This implies building out an internal practice to integrate into existing systems and processes to deliver the sought insights. In a field so wide, relatively recent and infamous for its buzzwords-per-second count, formalising problems and making explainable decisions is the only route that won't see you run out of resources and people's patience.
Metabase, a brilliant open-source BI tool, allows you to define your data model on top of the database schema, which includes descriptions, special types and table relationships. However, updating everything manually is not repeatable and prone to errors, so I will demonstrate how to export your existing dbt schema into Metabase automatically with a tool I created.
There comes a time in a software engineer's career when they start asking the dreaded question. Do I continue writing code as an 'individual contributor' or should I start the gradual descent into management?
Kotlin continues to conquer new areas of software development, but some are still firmly held by one or two languages. Data science, statistical analysis and machine learning are largely Python domains, but with many performance-focused implementations based on the JVM, Kotlin has a good chance to break into this scene too. The intention of this talk is to give a shallow overview of what you can do today.
Summary of the better talks, the more interesting themes, some conversations with other Kotliners and overall impressions about the first ever Kotlin conference in San Francisco. Think of it as your guide to what recorded talks to watch first.
Kotlin provides a mountain of features that Java developers previously never had access to. This creates endless opportunities. This also creates confusion akin to that of a kid in a candy store, which is exacerbated by the transition from simple beginner demos to production code, expected to be readable, performant and maintainable. Which to choose? Should I be doing this?