← Home
Learn Logs

Soak → apply, in public.

I pick something hard, go deep, and build with it until it sticks. Each log is one of those runs — start point, the work, the result. No polish, just the trail.

01
Pick
A hard thing with real stakes
02
Soak
Go deep, fast
03
Apply
Build something real with it
04
Show
Result, in the open

The logs

Shipped
Learning automation by building a triage line
Make.comClaude APIGoogle Sheets
PickI'd failed to "learn automation" from tutorials for years.
SoakBuilt a real pipeline with actual stakes instead of watching one.
ResultTimeverse — a working AI triage line for manual work. The build that proved I learn by doing, not watching.
See the build →
In progress
Cracking a competitive exam on a tight clock
12 July targetreconnaissance run
PickA timed competitive exam with a hard date.
SoakDesigning a prep system, not cramming — testing how fast I can build real recall.
ResultLogging the system and the score when it's done.
Planned
Pushing my chess rating with a system, not vibes
Chess.com: soumikcodes
PickA measurable skill with a public number attached.
SoakStructured study over random games. Documenting what actually moves the rating.
Planned
More AI tools, each one a new domain learned
PickA lead qualifier, a content generator, a chess coaching tool.
SoakEach build forces a new corner of the stack. That's the point.

More logs land as the runs finish. This page grows in public.

See the shipped builds → Why I work this way