Built By Sid
12 real-world mini products built around LeetCode algorithm patterns. Each one is a complete loop: spec, build, publish. No shortcuts on the hard parts.
The Why
I failed the last stage of a 6-stage interview because I didn't know algorithms well enough. That stung. But the industry is also changing — Google is moving toward AI-assisted interviews, and the question of what engineers actually need to know is genuinely open.
My take: we still need to problem solve organically. The shape of work is changing — we'll be building tiny systems, internal tooling, features of larger products — and for all of that we need to know when to prompt AI, when to write it ourselves, and how to judge the output.
So this project is my answer. I get to practice algorithms through building real products, and I get to ship things worth writing about. Kill two birds with one stone.
Algorithms through products
LeetCode patterns feel abstract in isolation. Building something real around them makes the pattern stick.
Honest constraint
The core algorithm gets built in a blank file with no AI. Everything else is fair game. The constraint is where the learning lives.
Build, learn, share
Every product gets a dev log on Substack. The writing forces clarity. The shipping forces completion.
Learning Go along the way
Most products are built in Go. Low supply chain risk, fast binaries, minimal deps. The language itself is part of the challenge.
The Catalogue
The Format
The Technique
What the algorithm is and why it's hard. No hand-waving.
Algorithm Ladder
V1, V2, V3. Complexity grows with the product. Ship early, improve often.
The Constraint
The rule you honour to make the learning stick. No AI on the core algorithm.
The Self-Test
How you know your implementation is actually correct. Every product has one.
Follow Along
Every product gets a write-up. The thinking, the mistakes, the algorithm, the shipping. No polish theatre — just honest documentation of the build.
Read on Substack ↗