I started planning for this project in February, began coding in April, first release in July, and been active daily till today.

At first I was only trying to make a tool that makes my day easier, I want AI to follow the exact procedure to perform my requests, soon I realize almost 80% of the "expected procedure" is deterministic without even needing AI, in most cases AI is only needed in judge/review/summarize/orchestrate actions, all the rest of the dirty work can be done without AI and without paying a dime.

As a result I created https://github.com/recued/recued, it stores all my data in my own home, executes predefined recipes whenever I ask for, fast, and cheaply.

Is it a success? I don't know, but I am still actively improving it on daily basis.

I guess neither success nor failure fits my description, I found something that I am passionated about, and it is also making my day easier.

From the sentimental perspective, I call it my success.

I might not even properly start until next year, but I've been working out bits and pieces for a video rendering system. The concept is similar to manim, but with much lighter-weight dependencies, and oriented around general text and diagrams rather than offering a bunch of special-case support for mathematical figures. I also plan to have a sort of DSL that can be used to describe video content in a highly readable way (meant to be rigidly structured, but accessible by non-technical people and also a useful target for LLMs), itself using a new markup language of my own design.

Specifically about dependencies: one piece of the puzzle is a lightweight wrapper for buffers (such as the builtin `array.array`) to allow treating them like Numpy arrays (still with accelerated code for simple operations like elementwise addition). Video basically requires a 4d-array (length x width x colour x time). For drawing, rather than PIL I'm working with Cairo and Harfbuzz directly.

Building a spaced repetition / incremental reading / video / podcast / audiobook app. I wanted a better UX version of SuperMemo basically. Something I actually enjoy using myself to explore topics incrementally. It uses the latest FSRS algorithm (FSRS-7), and also supports SM-20, SM-18, SM-2.

https://en.wikipedia.org/wiki/SuperMemo https://github.com/melpomenex/plethora

I'm building a data diode, with a pair of Raspberry Pi B+ computers for inside and outside proxy servers. I've got the first Pi booting, and am installing various Linux tools to get familiar with it.

I've been spouting off on the advantages of air gaps and data diodes from a theoretic perspective for more than a decade, I decided it's time to get my hands dirty.

I'll be open sourcing the whole thing, and writing code by hand in Free Pascal, the language I'm most familiar with.

i am creating a block universe of arithmetic truth

0101.0102[(0101)[0102

0103.0104[(0103)[0104

0105

[0101.0102 0101.0104 0102.0104

0204

in my formal calculus the above generates an addition structure relating arbitrary length bit patterns - because these are constraints, any known bits can be specified and the calculations can proceed backwards, forwards, or inside out

the formalism produces trees of parentheses by way of composition rules that are interpreted as in the Laws of Form "calling" and "crossing" (Spencer-Brown)

()()->()

(())->

the ultimate goal is to take high level programs and compile them directly into hardware descriptions without a simulation step

starting of the year I really wanted to make any video game , and after a long time of struggling with javascript , I ditched it , learnt c++ and finally this month I made a ghost game inspired by flappy bird starting from scratch :)

I've started planning the update of user-facing help for my 23 year old magnum opus. Help probably hasn't been updated in 10 years (I prefer coding) and I'm rather dreading the tedium. No, I won't be using LLMs - not in my lifetime - it's hand crafted all the way.

[flagged]