I have been working on several projects in my free time, the two most prominent ones are Harmonica and Gauss, below is an explanation as to what each one is about.
Harmonica is my project to improve on music self-hosting and algorithms my making a fully customisable and modular app, which can effectively run on anything, and host a music library.
I have applied my own knowledge of statistics (what I am planning to study in future), from the A Level syllabus as well as from textbooks and own research, in addition to research from a Large Language Model (namely the Bradley-Terry model for creating a larger ranking from binary inputs, which I have since researched extensively myself as it is very intriguing).
Now, I must state, a large portion of the code was written by a Large Language Model, however I am quite confident in its security, having run red team passes periodically, as well as keeping the code small and relatively verbose. The general input was mine. You can read the Large Language Model's account of the back and forth here.
Gauss is basically my attempt at making a harness for Artificial Intelligence, including but not limited to Large Language Models. The idea is to effectively take the best parts of what works (including websites, coding harnesses, and other applications), and implement what I would want to be using while not having the security risks and complicated irrelevant features.
I appreciate that it is not possible to effectively not have any complications, and it is a given that with a statistics-based model that something will go wrong at some point, but I hope to avoid that as much as possible.
Applications which this takes inspiration from include but are not limited to Odysseus, Openclaw, Claude Code, Codex, ChatGPT Claude Gemini Grok and Deepseek on the web, OpenCode, and many others (including forks of the above). The target is to currently get it self-sufficient enough and of good enough quality that it would realistically replace the current harnesses I use.
The vast majority of it is AI generated, including some decisions made, however I was still quite involved throughout making it.
I have not made this publically available as it is certainly not finished yet, but if you want to see it as it is today email me at contact@gauss.org.uk.
The reason why I chose this name was because Large Language Models are based on statistics (effectively what is the most likely response), and Gauss was a mathematician who studied statistics extensively, and the Gaussian distribution originates from his work. Another factor at play is that "Gauss" used to be part of a name I used, and still is in some places, so I was already ready to use it.
I am also working on this here personal website of mine, alexgeevs.com. The original version was just one plain html page I made based on my earlier knowledge of html, now I am working further using additional research. None of it is AI generated, this is all written entirely by me, both text and code. I plan to add CSS shortly once I figure out how it works.
Additionally I registered numerous emails under this domain using Microsoft Entra, and am effectively learning how to manage extensive digital infrastructure. If you are someone who would like assistance creating your own digital infrastructure, whether it be web domain, emails, or other, do let me know and I would be happy to help.
The last major project in this list is not yet put in motion. I am looking into making my own financial markets prediction model. I am aware that such programmes already exist (I believe they are named "Quant Models"), however I believe this is as good a time as ever to try, as currently it is an intriguing career path to me. I plan to create, partly from existing infrastructure and partly from scratch, a programme which would do its best to predict what a particular stock or bond would do tomorrow based on its past performance and numerous other factors. I appreciate this is unlikely to be successful (at least those I have spoken to seem to imply as such) but I am interested in trying despite that. If you believe you could be of assistance let me know, I would be very grateful
I have also worked on smaller projects within computers, including setting up an old laptop and external HDD as a rudimentary NAS, and also doing my best to effectively download my files emails and notebooks from my school email, which I managed to do quite effectively and can provide assistance if you have also finished Westminster School and would like to back up your data.