I’ve read about market microstructure for years — spreads, adverse selection, market making — but never built any of it. I wanted to see how far I could get using Claude Code as a Socratic tutor. The answer: further than I expected.
In the last few months, at least 8 people have asked me in surprise why I moved from pure machine learning into economics. Their reasoning was along the lines of why when AI is booming I am sidestepping and pivoting to something else.
My old NAS that I use for backups and store machine learning datasets I seldom use was becoming frustratingly slow. As a result I swapped out my Synology DS218j (with two 4TB WD Red drives), for a new DS423 with a third 4TB WD Red drive.
As someone who relies on multiple computers, I appreciate the convenience that personal cloud storage providers offer. However, I’m surprised by how uncompetitive the current offerings are. Dissatisfied with my Dropbox subscription, I started exploring alternatives and realized this issue is prevalent across the board.
About a week ago, Benjamin Zimmerman and I have released the “Synthetic Cancer - Augmenting Worms with LLMs” paper for which we have won the first place at the AI Safety Prize. Throughout the SCSD conference where we presented our work we have been asked a few questions by fellow researchers, journalists as well as the informed public.
As I will be going to the US for a few months this year, I knew I would not be able to take a huge Desktop PC with me. Hence, I started researching for solutions. The MacBook Pro 14 with M2 Pro seemed like a good fit.
As there is currently a heatwave rolling over Switzerland, I wanted to get detailed information on temperature and humidity. While researching some solutions to automatically record these values for multiple places wirelessly, I came across Ruuvi Tags. In the following tutorial I will outline the steps needed to create a Raspberry Pi based temperature, humidity, and air pressure recorder including a nice graphical front-end.
I used to have a Dell XPS 13 developer edition with a 4k screen. I was utterly disappointed with the 3-4h battery life and the weight of the laptop, hence once I saw a good offer for a used Lenovo X1 9th Gen, I couldn’t resist.
The following is an explanation how changes of bases work in vector spaces with a focus on building intuition.
Today we will take a glimpse at semaphores and have a look at an implementation based on locks. What is a semaphore? A semaphore is basically a lock that allows multiple processes to enter a critical section at the same time.
This document is a list of common sorting algorithms and their implementations in Java. The algorithms have been written by myself. I have tested them with the JUnit tests you can find at the end of this article, however, I do not guarantee that the implementations are 100% correct.
Cantor’s theorem states that for any set \(S\) we have \(S \preceq \mathcal{P}(S)\) and \(S \nsim \mathcal{P}(S)\). In words this means that the cardinality of \(\mathcal{P}(S)\) is strictly bigger than the cardinality of \(S\). Unlike some of my other posts, this one is strictly mathematical.
Pretty much every free newspaper you find comes with a sudoku puzzle. Depending on the difficulty it might take you a considerable amount of time to solve them. As a result I wanted to write an algorithm which could do that for me within seconds.
The goal of this short summary is to explain how you can build a small API based temperature sensor that costs less than 10$ and can be used anywhere. I use this sensor at my office to log the temperature and then display it on a smart dashboard.
Programs written in Go can be compiled for a variety of different targets. Cross compiling is easy, but takes a lot of effort. In order to automate cross compilation the following Ruby script is built to take that burden off you.