What am I working on ?

On my spare time...

---

I have started an IT personal project that takes most of my spare time, you can find some part of draft code here below this is the main tasks I have done, I'm working on or schedule to.
I also spend some of my time learning new things...

As I am diving more and more in details, and testing other programming langage, Golang and Rust mainly, instead of Python and Cython optimization, I'm finding more and more precisely what I need,
so I have started to rewrite and truncate unnecessary "testing/conceptual" code from my previous draft Python files and I'm converting some of those into Golang and Rust.

I have also made some simple tests with Rust, even if langage is not intuitive for me, dependancies management and compilation process seems very easy and reliable for the mains operating systems...
So I'm learning about it, and trying to see what are the strenght and weaknesses of those programming langage.
The goal is to make a kind of "roadmap" in order to stop all this searching and testing process, by defining what to code and with which programming language... As I cannot works on this project "full-time", I also started something simple to easily remember what I was doing or have planned to do... when I get back to this.

I'm starting to use AWS Lambda and test if it can be a good way to run scikit-learn with large datasets... for free. It also exist and "AWS Toolkit" extension for Visual Studio Code and it's my favorite IDE. Nowaday, many companies use AWS services (IAM, EC2, S3, RDS, Lambda), it's seems to be very interesting !
=> too costly in time, energy for learning and probably money at the end...

I'm testing and finishing my fastAPI server for real time technical indicator ; endpoint, function and parameters have been easely extracted from TA-lib package, but parameters have to be linked to 'data loaders' and implemented one by one...
=> will be transform, working on a main, home made and efficient Scheduler... (see 'scheduler' folder on my Github repo)
=> using of APS scheduler to run Python scripts (the most flexible and easy way found) => then precise statistical analysis server based on "hour of day" will be buildable...
=> I started to use ArticBD as main Database and will eventualy with Amazon S3 storage (I must look deeper about pricing and features)

I'm rewriting my custom 'Star Queue' and components linked with asyncio (far more efficient)
My custom 'Star Queue' is a, let's says, mix of 'Star', 'Mesh' and 'Extended Star' used throught SSH tunnels and local port forwarding:

I have spent many time writing code on the same type of things, and thanks to "AI" coding assistant, I figure out that switching to Golang langage was not so difficult than I tought, even if it takes time...

So finally I decide to use not only Python and Golang but also C/C++ or Rust and probably Java, for what they are best at.

I have spent many times searching, trying to code and test for standard or tricky technics to share data efficiently with lightweight containers and low/ultra low latency communication between different processes on the same machine. (Memory Map, *Files descriptors, Kqueue-Epoll-IOCP, 'Zero Copy' serialization, libuv, syscall, OS signals, jump tables...).

In my play lists of videos specially talking about the topics of "programming and trading", I heard talking about this concept : "Mechanical Sympathy" and "Aeron: Open-source high-performance messaging" which, I hope will does exactly what I want/search and even more ! => will be coded in Rust .
So I'm learning and testing Java, it's wide ecosystem and above all Aeron, in order to integrate it with as my pub/sub message processing for trading. => not needed anymore...

I will start to play with Black-Sholes model to understand deeply option market pricing and trying to spot "gamma squeeze" programmatically...
Maybe a create a simulated bond calculation...
Create a union file system module for a preconfigured Postgre server
Make global "correlation analyst" (pearson correlation)
Check if crypto broker plateform overload (error message when occurs) is exploitable (a "pseudo arbitre" is also running on this machine)