Cerebrum
Following a multi-service SaaS from architecture and tenant isolation through detection mathematics, dashboards, threat intelligence and operational hardening.
See the build trail →The public workshopI make too many things.
I make too many things.
Here is the evidence.
IamANeRD.xyz is where finished products stop pretending they appeared fully formed. It documents experiments, architecture choices, failures, rebuilds, commands, book work and the useful lessons hiding inside the mess.
42-ishthings being created at once
Real notesdecisions, mistakes and fixes included
No gatekeepinglearn what the command actually means
#
docker composeBooksgit statusWhy?On the bench
Current work, with the casing removed.
The point is not to perform expertise. It is to preserve the thinking: why a choice was made, what broke, which file became authoritative and how the next person can continue without reconstructing the entire history from fragments.
Git for AI creation
An application concept for recording chats, prompts, decisions, artifacts, branches, provenance and the current source of truth across dozens of simultaneous creations.
Inspect the concept →Books in public
Chapter architecture, revision systems, production decisions, companion files and the strange discipline of turning technical knowledge into readable books.
Read the working notes →How the workshop works
Show the decisions, not only the result.
Most project histories preserve the final code and lose the reasoning. This workshop treats context as a first-class artifact.
Start with the real problem
Write down what is painful, what keeps being lost and what “better” would mean in practice.
Freeze important decisions
Names, contracts, architecture boundaries and product principles become explicit instead of quietly drifting.
Build in recoverable steps
Backups, full-path commands, whole-file replacements, validation and rollback are part of the work—not cleanup after it.
Publish what was learned
A failure that is explained becomes reusable knowledge. A success without context becomes a future mystery.
Useful starting points
Take something back to your own terminal.
Tutorials focus on repeatable workflows: safe deployment, Docker and Traefik, WSL and Linux fundamentals, debugging, backups, context transfer and the boring details that prevent dramatic evenings.
Learn through WSL
Use Windows Subsystem for Linux as a safe place to learn commands and concepts that apply far beyond WSL.
Deploy without collateral damage
Inspect first, back up, change one bounded system, validate and know exactly how to reverse it.
Preserve context
Carry decisions and source-of-truth files forward instead of hoping a future chat reconstructs them correctly.
Write for the next tired human
Full paths, explicit assumptions and commands that explain themselves beat heroic memory every time.
There is always another project.
The workshop will keep expanding as software, books, websites and AI experiments move from idea to working system.