About

Hello, I am Kalun.

I work in AI and make videos about what I see happening in technology, and why. This page is the short version of who I am and how the videos are made.

Portrait of Dr. Kalun Ho

Who I am

Head of an AI Lab, doctorate in computer vision.

I lead the AI Lab of a large German company. Before that I headed IT data science there, and I have been with the company since 2010, when I started as a student. My doctorate is in computer science, with a focus on computer vision, from the University of Siegen.

My day-to-day work is practical: running on-premise GPU infrastructure, making language models usable inside the company, and building agents that take over routine processes. I also give talks on AI at company events, universities and career fairs, in English and German.

Why the videos

Most of it can be explained in plain terms.

In my work I explain things like transformers or computer vision to managers and engineers, and the technical part is rarely the problem. It just needs someone to take the time. The videos are my attempt to do the same in public, one question at a time.

They are a personal project and do not speak for my employer.

What comes with each video

A handout and a deck, both free.

  • The handout. A free PDF with the charts and sources, linked in the video description and on the videos page.
  • The deck. The editable PowerPoint version, linked next to the handout. Use the slides in your own meetings. Credit is appreciated, not required.

How the videos are made

Research first, then the script.

Each video starts with the research. Figures are cited and listed in the handout, charts are redrawn from the source data, and illustrative examples are labelled as such. Where the evidence is thin, I say so.

Where no source has the number, I measure it myself. Each of those experiments has its own page with the setup, the full results and the code.

What stays off the channel

A few things you will not find here.

  • Investment advice. Nothing here is a recommendation to buy or sell a security.
  • Confidential material. No confidential information from my employer or from clients.
  • Speculation about people. I do not guess at the motives or plans of named individuals.

Contact

Questions and corrections are welcome.

Ask in the comments under the video, where others see the answer too. For an experiment, open an issue in its repository on GitHub. For anything else, write to me on LinkedIn.

If a figure looks wrong, name the video and the source number, so I can check it against the source.

See the next video when it is published.

Subscribe on YouTube. Each video links to its free handout and deck in the description.