Why I Started Workbench

Updated

Workbench is my little corner of the internet where I post about what I am learning and building in AI and data science.

I'm a statistician by education, a data scientist by profession, and an avid learner of new things.

My current rabbit hole is AI. Specifically agentic AI.

My process for learning new things

I learn best by doing.

I retain next to nothing by reading articles, tutorials, or long-winded guides. And I definitely do not retain information from listening to podcasts or watching YouTube videos (the amount of rewinding I do in either of these formats is embarrassing).

But I still do it partly because it gives the illusion of mastering a topic - I think that if I can just read one more really in-depth tutorial then it'll finally all click for me. But it never does. Plus I'm quite lazy and impatient - I want to learn the thing yesterday.

The only way I effectively learn new things is to do the thing.

Programming: build some cool projects; homelabbing: self-host something on a Raspberry Pi and access it on all of my devices; agentic AI: build a fully walled-off local AI system that automatically picks up tasks I assign it through a self-hosted Vikunja instance (maybe a little overkill for learning, but whatever); and so on (you get the idea).

However, I cannot (and should not) reinvent the wheel whenever I want to understand how the wheel works. So the choice of what to build can often be a tricky one. When starting out, I give myself permission to chase as many shiny objects as I want. The point is to get excited about what you're building and learning. That way, you're more likely to follow a project to the end.

I'm sure a lot of people are in the same boat as me. In fact, one of the most common pieces of advice you can get when asking how to learn something new (and I'm talking specifically in tech) is to learn by doing.

There are a lot of creators out there building pretty incredible things. But so many of them hide it all behind a paywall of some kind.

So I'm starting Workbench to share the ins-and-outs of the things I build. I'll include my mistakes, lessons, notes. The idea is to give a behind-the-scenes look into my thought process, not just a README with a summary of the final_final_final version.

Some stuff is built by and with AI. Some of it is hand-coded by me only. However, everything on Workbench is written entirely by me. This is my guarantee to you. There will be no AI slop found here. Just my messy, sometimes incoherent thoughts.

What you can expect from me

On a regular basis you can expect:

You can also find information on my hardware, the software I'm running, and how my local AI system is built in The Setup.

Stick around. Read a little. Steal some of my ideas. Implement them yourself.

Learn, build, grow.