Of artificial intelligence and rockets at Mach .5: What I’m currently working on

Of frustration, Pytorch and long waiting times

I’ve always been fascinated by AI projects like GPT-3 (Generative Pre-trained Transformer 3) or Quick, Draw! (an experiment by Google in which a neural network tries to recognize users’ drawings). That’s why I’ve been working on a project about artificial intelligence for quite a while now. I don’t want to spoil too much yet, but this much can already be revealed: 20,000 tweets, extracted from Twitter over seven long hours, fed into a program library ("Pytorch"). The fact that downloading 20,000 tweets from Twitter takes so long is due to the very frustrating Twitter API version 2, with which tweets can only be downloaded in very small numbers.

More updates on the project soon!

Micro rockets with maximum performance

Over the last few months, I’ve also been diving deeper into rocket design and rocket construction on the side (both, of course, on a model scale). Right now I’m trying to make the electronics for determining altitude and acceleration as small as possible — my goal in this project is to build small rockets with (comparatively) powerful rocket motors. The rocket motors make up more than 4/5 of the rocket’s weight, so altitudes of 400–600 meters should be possible. Since the rule of thumb that rockets with less mass accelerate higher initially also applies at these small scales, I’m expecting accelerations of up to 60 g, that is, 100 times Earth’s gravity (that means reaching 100 km/h in 0.05 seconds; humans can survive around 5–8 g briefly).

Project Point

A project that is currently taking up a disproportionate amount of time, but about which I don’t want to reveal anything yet, has also been keeping me busy for a few weeks now. Just this much: It’s something completely different, heading in another direction from my previous ventures, but I’m still really looking forward to it.