Why I'm writing in public — John Twipraham Debbarma

I keep a private notebook of every idea, derivation, bug-hunt and "wait — why does this work?" moment that crosses my desk. The notebook is messy on purpose; it's where I think. What I'm starting here is the public version of the same impulse — concept posts, not announcements.

Why publish

Three reasons, in increasing order of selfishness:

Forces clarity. Anything I can't explain in a couple of paragraphs I don't actually understand. The act of writing for someone else is the cheapest way I've found to surface gaps in my own model.
Cheap durable artifact. A blog post takes an hour to write and lives forever. A conversation I had in lab last week is already half-faded.
It's a slow signal to recruiters. If you're reading this and we haven't met — this is who you're hiring. The portfolio shows what I've shipped; the blog shows how I think.

Shape and cadence

Concept-first. A post anchors on one idea: a derivation, a technique, an algorithm, a misconception I had to undo. Not project announcements, not life updates.
Daily or weekly, depending on what's in front of me. Some weeks I'll publish three posts in two days because I'm deep in something; some weeks none, because I'm busy and shipping is more important than writing about shipping.
Short by default. 600–1500 words. If it doesn't fit, I'll split.
Code where code is the point. Math and diagrams where those are.

What's queued

Half-drafted in the notebook, in no particular order:

The optics behind the Liquid Glass refraction filter I wrote for this site, and why scale ≈ 9 falls out of crown-glass refractive index.
Sensor-fusion intuitions: when LiDAR and mmWave radar disagree, who do you believe?
A clean derivation of PCA from "I want to project data into a lower dimension and lose the least variance".
Why Ramsey's theorem makes the Sim game on K₆ undrawable, in a way you can feel.
Notes from seven months in a defence R&D lab — what I can say, what I can't, and why the distinction matters.

If any of those sound interesting, the contact page is the fastest way to reach me with a "write that one next."

Why publish

Three reasons, in increasing order of selfishness:

Forces clarity. Anything I can't explain in a couple of paragraphs I don't actually understand. The act of writing for someone else is the cheapest way I've found to surface gaps in my own model.

Cheap durable artifact. A blog post takes an hour to write and lives forever. A conversation I had in lab last week is already half-faded.

It's a slow signal to recruiters. If you're reading this and we haven't met — this is who you're hiring. The portfolio shows what I've shipped; the blog shows how I think.

Shape and cadence

Concept-first. A post anchors on one idea: a derivation, a technique, an algorithm, a misconception I had to undo. Not project announcements, not life updates.

Daily or weekly, depending on what's in front of me. Some weeks I'll publish three posts in two days because I'm deep in something; some weeks none, because I'm busy and shipping is more important than writing about shipping.

Short by default. 600–1500 words. If it doesn't fit, I'll split.

Code where code is the point. Math and diagrams where those are.

What's queued

Half-drafted in the notebook, in no particular order:

The optics behind the Liquid Glass refraction filter I wrote for this site, and why scale ≈ 9 falls out of crown-glass refractive index.

Sensor-fusion intuitions: when LiDAR and mmWave radar disagree, who do you believe?

A clean derivation of PCA from "I want to project data into a lower dimension and lose the least variance".

Why Ramsey's theorem makes the Sim game on K₆ undrawable, in a way you can feel.

Notes from seven months in a defence R&D lab — what I can say, what I can't, and why the distinction matters.

If any of those sound interesting, the contact page is the fastest way to reach me with a "write that one next."