Jeremy Elbourn
jelbourn.bsky.social
Jeremy Elbourn
@jelbourn.bsky.social
Software engineer @ Google, tech lead for Angular. Physically in Seattle, mentally in Faerun.

Talk to me about accessibility, code health, ui components, D&D, indie games, scfi-fi + fantasy, woodworking
Overall this is a cool and ambitious comparison!

Happy to answer any questions
October 27, 2025 at 8:43 PM
The `@import` for charts.css inlines the styles into the components, which ends up putting them in a string in the JS bundle. For the comparison, it might be better to import chart.css in a global stylesheet.
October 27, 2025 at 8:43 PM
Hey! A couple of small notes about the Angular app

It's pulling in FormsModule for ngModel and ngSubmit without using the rest of the pkg. It would shave off a fair bit of JS to handle the form submit like the React app does (especially since the other frameworks don't have a comparable forms pkg)
October 27, 2025 at 8:43 PM
I've been using a Herman Miller Mira 2 for nearly a decade and it's been great. The color availability was also a big selling point.
October 22, 2025 at 8:36 PM
IMO Hades is best for a trip, since it's less mental overhead. I also think Hollow Knight is an overall stronger game that Nine Sols (which is still good) for their shared genre.
October 3, 2025 at 7:21 PM
It's always satisfying to let out your little automaton before the fight so the boss gets poisoned while they're vogueing.
September 22, 2025 at 6:44 PM
So try out the tool and tell us what you think!
September 16, 2025 at 4:42 PM
We also identified a few places where we could make fixes in the framework itself for patterns the LLM was producing that didn't work, but probably *should* have. For example, we made binding ARIA attributes in Angular more flexible and intuitive.
September 16, 2025 at 4:42 PM
The tool also really helps debugging issues where LLMs commonly go wrong. By iterating with this tool, we were able to fine-tune our best-practices prompt on angular.dev/ai/develop-w... and watch the score go up.
September 16, 2025 at 4:42 PM
So the team built our own tool so that we can analyze and iterate on LLM codegen quality in an evidence-driven way. We still have a lot we want to add, but even today it's a great way to measure the quality of web code coming out of LLMs in a consistent, repeatable way.
September 16, 2025 at 4:42 PM
It's been clear that more and more developers are reaching for LLM tools to help author code. On Angular, we were encountering a lot of claims around the relatively quality of different models, tools, frameworks, etc. But nobody was really measuring, or measuring the things that we cared about.
September 16, 2025 at 4:42 PM
It helps if you imagine the run back as phase one of the boss
September 16, 2025 at 4:47 AM
I love the Bob's Red Mill no-sugar-added granola.
August 27, 2025 at 3:47 PM