Mix · Pascal van Kooten

htop.astro

I have had a few trending posts in the past, here I show them using htop:

  1  [|||||||||||||||             54.2%]   Tasks: 1337, 42 thr; 1337 running
Mem  [|||||                       19.5%]   Uptime:   36 years, 314 days, 18:23:42

  PID USER   VOTES  %CPU %MEM   TIME  COMMAND
  542 pascal   542  17.8  4.1   9:14  brightness-ml.py --on reddit
  510 pascal   510  16.7  3.7   6:52  whereami --predict --on reddit
  230 pascal   230   7.5  1.8   4:02  yagmail.send_email --on reddit
  200 pascal   200   6.6  1.2   4:18  codingame_ai --top5 --on reddit
  170 pascal   170   5.6  8.7   3:31  neural-complete --train --on hackernews

F1Help F2Setup F3Search F4Filter F5Tree F6SortBy F9Kill F10QuitKILL 15SIGTERM 9SIGKILL · Esccancel

Avatar.astro

Giscus.astro

Comments slot

Ticker.astro

PSCL ▲ +2.34 $42.07 strong commit volume · [INFO] git push origin master · 30 files changed, +127 -42 · ◇ LLM throughput up 18% q/q · vibe coding cited as leading indicator · WHRMI ▲ +0.81 $18.40 indoor positioning still works · [INFO] commit 8978: "fix easter egg" · BRGHT ▼ -0.05 $9.12 ambient light low · ML model pre-dimming · HFT ▼ -0.42 $99.99 proprietary · cannot disclose · PSCL ▲ +2.34 $42.07 strong commit volume · [INFO] git push origin master · 30 files changed, +127 -42 · ◇ LLM throughput up 18% q/q · vibe coding cited as leading indicator · WHRMI ▲ +0.81 $18.40 indoor positioning still works · [INFO] commit 8978: "fix easter egg" · BRGHT ▼ -0.05 $9.12 ambient light low · ML model pre-dimming · HFT ▼ -0.42 $99.99 proprietary · cannot disclose ·

Footer.astro

Chat.astro

Navbar.astro

Introduction.astro

introduction.md ×

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

I hold an MSc in Methods & Statistics. My early days were two years on
blockchain, then eight years across data science and software engineering,
a chapter building chatbots just before LLMs landed, and finally several
years as a quant developer in HFT — which means everything I wrote
during that stretch was proprietary, and open source took a back seat.
I left the industry at the worst possible time, right as GPTs came to
be. I am back at it now, with renewed energy and sharper tools.

I am highly competitive and used to pour that into online machine
learning and AI competitions. Around the same time I became an active
open source contributor and spoke publicly about machine learning —
and innovation more broadly — on a number of occasions.

What I love most is empowering humans with machine capabilities. I am
currently running my own company.

I no longer take on freelance work. These days I am open to whatever
interesting projects happen to cross my path, and pick them up on
their own merits.

I primarily write Python and Rust, and let LLMs handle the occasional
foray into C, HTML, or JavaScript. I enjoy vibe coding, but care just
as much about using AI to ship genuinely high-quality output.

My editor is Emacs — although, who really uses an editor anymore for
coding?

Ironically, I was a generative-AI skeptic for a long time — I simply
could not imagine anyone breaking copyright law at the scale required
to train these models. Reality had other plans.

⎇ main UTF-8 LF Ln 1, Col 1 Spaces: 2 Markdown

SpecialFX.astro

CLI.astro

guest@vks.ai — ~

▶_ Terminal — guest@vks.ai

guest@vks.ai: ~

9:41 􀙇

9:41 ▴ ▼

Terminal ⋮

█

speaking-at-globe-tokyo.png

fouR.gif