(c) 2025 Kilian Singer
These little light tutorials are ment to show important tricks with programming and tools to be efficient in STEM (science, technology, engineering and mathamatics). This includes how to use cutting edge ai tools for programming, Retrieval-Augmented Generation, LLM finetuning, OCR for even math, and more. Learn how to self-host a server from home and use most advanced commercial computer algebra systems for free.
Note
Instead of browsing these files through the web I recommend going through the first >part and then browsing it through visual studio code. As a starter you can download >the first two parts here: codetricks
Then open the first part and start your journey...
02 - Revision Control with GIT
04b - My webpage on raspberry pi
08 - Literature Research with AI
P2.3 Constants, References and strings
P3.3 Assignment operators and miscelanousoperators.html
P4.1 Loops and Jumps - Keyboard.html
P4.2-Conditions and Loops.html
In these chapters we will use our new skills to finetune large language models to specific tasks. I will also show you how to finetune Optical-Character-Recognition such that it can read and convert your handwritten lecture notes including math formulas into markdown, even if you have a terrible handwriting. Also I found the first speech recognition software that can transcribe my lecture with all the mixture of german, physics, english. I will show you how to get aroundt the memory demand that the original code had by segmenting your task. These are currently all highly sought programming tasks in companies. So have a look and add it to your skill portfolio.
10 - Finetuning Large Language Models
11 - Cutting edge Optical-Character-Recognition
12 - World's best speech recognition for free
If you want to dive into the basics of neural networks have a look here (Niesen also coauthored a book about Quantum Information Theory...) http://neuralnetworksanddeeplearning.com/
Tip
Choose a small programming project > that challanges you a bit. This is > a learning paradigm called learning the hard way. Which basically means that you learn best by trying a problem which is not too easy. Many years ago I needed to learn about compiler design, that's when I realized that there is no better way to learning all the nits and bits of a programming language and also value its design descissions and beauty by writing a compiler yourself.