vol. iii · ongoing

The Notebook
of yash tyagi

software engineer by profession.
writer, note-taker, doodler, and observer by habit.

p. 01

Hi, I'm Yash.

I build software whenever something annoys me enough that I can't stop thinking about it. Most of my projects start as messy sketches in a notebook long before they become code.

I spend a lot of time wondering why some products become part of people's daily lives while others get forgotten after a week. That's probably why I keep redesigning things in my head even when nobody asked me to.

I collect ideas faster than I can build them, so I keep a "morgue of ideas" where unfinished projects wait for the right time to come back to life. I doodle on almost every page, rarely finish every idea I start, and keep building things simply because I like it.

a small confession

I notice small frustrations the way other people notice the weather. It's not a discipline. It's a condition.

— 01 —

p. 02

how it actually goes

the only honest flow chart I've ever drawn:

Problem
Get curious
Research
Write everything in notebook
Draw user flows
Design
Build
Get distracted by 3 new ideas
Write those down too

— 02 —

What I'm building right now

Bookswagon — Flutter app, end to end

shipped · 17× orders

I rebuilt Bookswagon's mobile app from scratch — alone, end-to-end. The existing app was a .NET WebView wrapper averaging 3.4 orders a day with a Play Store rating near 1.8. I owned everything from Figma to Flutter to staged rollouts.

In 8 weeks: 3 → 58 orders/day, MAU 4.7K → 9.4K, Day-7 retention 3.2×, crash rate <0.04%.

The numbers weren't the achievement. The achievement was knowing real parents were now having a smooth experience where before they had frustration.

fig 1. first sketches — 4-5 buttons, no more

fig 2. the federation graph (in progress)

Next.js + GraphQL Federation BFF

in progress

Migrating the legacy web stack to Next.js (SSR/SSG) and designing a federated GraphQL layer over Orders, Cart, Payments, Inventory, and User services. The goal: one schema, many clients, none of them lying to each other.

the pinboard

Confidence was cash. You had to have some to get some.

— Phil Knight, Shoe Dog

idea: a reader that hides the title until you finish.

Until the lions have their historians, the history of the hunt shall always glorify the hunter.

— Chinua Achebe

morgue of ideas

FlowForge — make a chat conversation portable across models.

the metric that mattered

p. 03

orders / day

58+17×

was 3.4

mau

9.4K+2×

in one month

day-7 retention

3.2×

onboarding rewrite

crash rate

<0.04%

staged rollouts saved us

checkout funnel — before

Product View100%

Add to Cart35%

Checkout18%

Order Complete11%

↑ the 65% drop. that was the whole story.

we kept asking users for things they didn't care about

— lesson, written in pencil

The drop wasn't because users disliked books.
It was friction.
Users rarely leave because of one giant problem. They leave because of five small annoyances.

paper — a writing app

v0.4 · weekends only

A small editor for people who think on the page. No AI. No collaboration. A blinking cursor and good typography. Started as a weekend rewrite of an old prototype. two three rewrites in.

~/paper $ git log --oneline -5
a3f9c1 trim the empty state to one sentence
812bd0 stop fighting the caret, let it blink
9f4e2a delete the toolbar (again)
1c0a7b move autosave to the margin
7e1d44 first draft of the writing surface

commit a3f9c1d

Author: yash

Date: Tue Jun 4 09:14

trim the empty state to one sentence.
// a blank page should feel like permission, not pressure.

People rarely adopt a product because of what it does. They adopt it because of what it lets them become.

— margin note, No Logo· naomi klein

currently reading

A strange mix, on purpose.

No Logo

Naomi Klein

Shoe Dog

Phil Knight

Bravehearts of Bharat

Vikram Sampath