PUBLIC ARCHIVE · BUILT IN THE OPEN

Project Archive

Every build gets a lab note — real code, real data, read it end to end. From the autograd engine and the code-reading TUI to the daily AGI research log. All of it on GitHub, all of it open.

—Total builds
—Shipped live
—Open source
—In progress
01 — Full list

All builds

Loading archive…
01Daily log · in progressCode

AGI_Research

Public lab notebook — committed daily since September 2026. Linear algebra, calculus and probability from axioms → neural nets and backprop in pure NumPy (SGD through Adam) → now CNNs, Transformers and PyTorch. No shortcuts.

backprop ✓NumPyCNNsTransformersPyTorchdaily commits
Jupyter Notebook · Python
02On PyPI · shippedCode

micrograd_hk

A tiny autograd engine on NumPy — a Value graph with vectorised backprop and gradient descent, written before touching a framework. Inspired by micrograd, but matrix-driven: pip install micrograd_hk, Colab demo included.

autogradNumPybackpropAdam-MPyPIMIT
Jupyter Notebook · Python
03On PyPI · shippedCode

peek

htop for codebases — understand any repo in five seconds. Languages, size and structure in one screen: pip install peek-code && peek .

CLITUIRich + TextualPythonPyPIdeveloper tools
Python · HTML
04v1.1.0 · shippedCode

inkdown

Notion for Markdown — write in a clean editor, ship plain .md files. Local-first desktop app: no account, no lock-in.

MarkdowneditordesktopPyWebViewlocal-firstMIT
JavaScript · CSS · HTML · Python
05Open sourceCode

practice_code

60+ Python solutions for algorithms and data structures, grouped by topic. Each notes time and space. Fork it, add your approach, open a PR.

algorithmsdata structuresPythonDSA
Python
06Open sourceCode

pkg_jupy

Jupyter experiments that grew up — small Python packages, utility notebooks and visualisation scratches. Open for issues and PRs.

Jupyterpackagesutilitiesexperiments
Jupyter Notebook
02 — Next

On the bench

the archive grows
In the lab · coming soon
Next: Sentiment

Every project here gets a lab note — real code, real data, runnable in your browser. The next bench is already training. Follow commits on GitHub or check back.