Laurium Labs attended JuliaCon 2019 in Baltimore, Maryland. Julia is a programming language that aims to be a "fresh approach to technical computing". Inspired by the so-called "2 language problem", Julia aims to be both easy to read like Python and performant like C/C++. Many popular Python packages have C/C++ under the hood (resulting in a tiny number of actual maintainers), and the overhead of switching between languages is enough of a burden that Julia was created.
Here at Laurium Labs, we prefer using Julia for our scientific and data science solutions. Attending JuliaCon gave us the chance to meet some of the authors of the language and hear the buzz first-hand. For example, Mark was doing some data science work, and upon updating a package, he saw a bunch of deprecation warnings. Mark looked across the room and spotted the author of the package, so he ventured over and showed the warnings to the author (Jacob Quinn), ascertaining if he should bother to update his code to get rid of the warnings. It really is interesting to talk to the authors behind the packages you use on a daily basis, face to face interaction can't be replicated through a screen.
Conferences really bring together a very like-minded set of people. You can go up to almost anyone and discuss a very niche topic, programming in Julia. Mark opened many of his conversations with "So how do you use Julia?" and the answers always sparked an interesting conversation. While being a very niche topic, conference attendees ranged from undergrad students to semi-retired industry legends. The breadth of the people and their knowledge was tremendous.
Laurium Labs is currently working on a mathematical optimization project using the JuMP package, and we learned plenty about potential approaches and pitfalls we should be on the lookout for. Many people were excited to hear that Laurium Labs was using Julia for industrial applications on the shop floor. Brent's loud voice meant that it was hard to miss hearing about it if you were in the general vincinity.
Some of the highlights for us of the conference included Chris Rackauckas's talk on differential equations and machine learning, which was a downright inspiring, mile a minute talk: Scientific AI: Domain Models with Integrated Machine Learning. Stefan Karpinski's talk on on multiple dispatch made a strong case for multiple dispatch (embodied in Julia) to be adopted as a widespread programming technique. The helpfulness at JuliaCon is amazing, you can get enthusiastic expert advice on your problem in an instant. This alone is worth the ticket.
Seeing the great work by others at JuliaCon has us ready to roll up our sleeves. We hope that Laurium Labs will present some real-world Julia solutions at JuliaCon 2020, perhaps a JuMP solution and a domain model integrated with machine learning will be presented!