I'm Kilian (lusoris), based in Germany. I have built software with AI coding agents for about a year, with many good tips from Lawrence Curtis. I value precision, clear rules and evidence.
- Consistent video scores. I want video quality measurements to agree across CPUs and GPUs so encoder tuning rests on evidence.
- AI on less power. I want to explore biologically inspired AI that works within small energy budgets.
- Accountable coding agents. I want written engineering rules checked by machine through Praetor.
- Reusable software. I want a small ecosystem of composable Go and SvelteKit modules.
- Details matter. I treat small inconsistencies as problems worth fixing.
- Rules apply to me. I document standards and automate their checks for my own work.
- Understand first. I study prior attempts, weigh costs and revisit choices when evidence changes.
- Measure the result. I prefer small solutions whose costs and results I can measure.
- Contribute upstream. I send fixes back to their projects; 27 of my pull requests are merged across 10 projects.
Fixes that left my own trees and were merged where they belong:
| Project | Merged |
|---|---|
| FFmpeg | 16 |
| OpenVINO Model Server | 2 |
| Vulkan Validation Layers | 1 |
| Paperclip | 1 |
| Paperclip Aperture | 1 |
| Hindsight | 1 |
| helm-charts | 1 |
| readarr-rresurrected | 1 |
| configarr | 2 |
| argo-cd | 1 |
You can sponsor my work through GitHub Sponsors. Sponsorship buys no support and no priority.
My avatar is a character in the style of the Go gopher. The Go gopher was designed by Renee French and is licensed under the Creative Commons Attribution 4.0 licence. The upstream section is generated from upstream.json by tools/render.py.