Watching this video from Daniel Leivers demonstrating his master’s dissertation this week set me thinking.

What Daniel has built is clearly impressive. You can sketch UI with your iPad’s pencil and see real interactive UI or Swift code appear alongside it. The app uses a model created from sketched UI components fed into the yolov5 algorithm and converted to a CoreML model to generate code from the identified UI components. It’s a good idea and makes a fantastic demo. If you’re interested in reading his dissertation, it’s available here.

I took a couple of things away from Daniel’s video. With all the light shining on GPT and LLMs at the moment, it’s easy to forget that CoreML is still capable of building amazing features like this. It’s also easy to think that AI means “running in the cloud” with how resource-intensive LLMs are, but this all runs locally.

But more significantly, his video set me thinking about progress. Thinking back to my own final year project that made up part of my software engineering degree that I completed in 1996, 28 years ago. I created an HTML editor using Borland Delphi 1. Remembering what my app did, it seems almost comical. It did not have an integrated browser preview, syntax highlighting, auto-completion, auto-indentation, or any number of other features that would be table stakes for even the most basic code editor today. To give you an idea, it was a slightly less capable version of something like this. I was still proud to complete it, though, as I am sure Daniel is proud of his project.

But still, I stood on the shoulders of giants to create my little 16-bit text editor, and so did Daniel with his remarkable iPad app. There are obvious dependencies like the machine learning algorithms and CoreML, but there are so many more that you might not immediately consider. Think about some of the more “mundane” parts of his app. How difficult would it have been to implement the sketching functionality without PencilKit? Not to mention technologies like SwiftUI, UIKit, the Swift language, Foundation, and even Darwin!

I don’t say this to undermine the amount of work Daniel did on his project. I want to celebrate the amazing technologies and development environments we have access to today that make projects like this possible as dissertations.

We truly do stand on the shoulders of giants I stood on the shoulders of the giants that created Borland Delphi, HTML, and the web back in 1994 and Daniel stands on countless researchers and computer scientists that made his project possible. One of the things I love about this industry is that even if we build an app in a team of one, we’re not working alone. Whether open or closed source, the APIs that others create are constantly pushing the industry forward and it’s a huge part of why it’s one of the fastest-paced industries that exist. I find it incredibly inspiring.

Dave Verwer  





iOS Engineer @ trivago – trivago, a metasearch engine using real-time auction and petabytes of data, enables millions of travelers compare hotel prices from hundreds of booking sites. Based in Düsseldorf, we foster a culture of learning and innovation, embracing flexibility for our talents to shape the travel industry. – On-site (Germany) with some remote work (Anywhere)

Senior iOS/macOS Developer @ Paste – Joining Paste means crafting impactful, user-focused products alongside a team that values innovation, flexibility, and a culture of collaboration. Dive into projects that push the boundaries, enjoy freedom in how you work, and help shape the future of productivity tools. – Remote (within European timezones)


If you’d like to see your job featured in iOS Dev Weekly, post it on iOS Dev Jobs and select “Featured listing” as you check out, and it’ll be in next week’s newsletter. 🎉


And finally...

You thought you had a tough debugging problem? Think again!