last ink: Oct 28

Abdelrahman Amgad

Abdelrahman Amgad
hey there o/

Software engineer. I take vague problems from the first customer call all the way to production.

NowOn Incorta's Service Engineering team. Final year of Computer Engineering at Alexandria University, done June 2027.

Proof

Abdelrahman Amgad
hey there o/
Cinemate before, as a rough sketch: every post waits about two seconds on an AI model, and the data is spread over three databases. postmodel2 s ??? Cinemate after, drawn precisely: you reach a single gateway, which leads to the backend and the watch-party service. The backend saves a post in 10 ms to one database and puts it on a queue, where models judge it and send the verdict back in 0.7 seconds, with no requests lost. yougatewaybackendwatch partydatabasequeuemodelmodelmodelsaved10 msverdict0.7 sthe only door in.3 → 1everyone stays in sync0 requests lost

↑ every project starts on the left.at the top.

fig. 1: Cinemate, before and after I rebuilt it.the whole story →

it's 3am. you're here.
I'm probably awake too.

Work

Things I built, start to finish.

Notebook

Experiments, and what they taught me.

  1. Apr 2026

    Unplugging Cassandra nodes to see what still answers

    Three nodes, three replication settings, every combination of failures. A table of exactly when "consistent" stops being "available."
    experiment
  2. Apr 2025

    Teaching a tiny kernel to share fairly

    Inside PintOS, a teaching operating system. Fixing what happens when a low-priority thread holds the lock a high-priority one needs.
    coursework

Principles

How I work.

Written down, so I can be held to them. Each one points at the proof.

If I own it, I own all of it.

From the first messy meeting to the thing running in production. Nobody should have to check on it. Like this →

Half-baked is worse than not started.

If I put my name on something, it gets everything I have. Otherwise I say no early. Cinemate kept going after the grade was in →

learned this one the hard way.

Make the call. Write down why.

Decisions are cheap to make and expensive to forget. Every fork gets a line in the log, so future me can argue with past me. Here’s a log →

Curiosity is a work ethic.

The best ideas came from something I didn’t need to know yet. Like unplugging databases to see what still answers →

Build for the human on the other end.

It isn’t done when it works. It’s done when it feels obvious to the person using it. A fire alarm that needs a server arrives late →

Leave people better than I found them.

The best thing I can give a team is people who don’t need me anymore. Mind Cloud →

the hard part is letting go :D

Have something worth building?

An app you need built, a team one builder short, or a half-formed idea that needs someone who finishes things. Reach out.

built by hand, mostly at night.↑ Back to top

nobody really knows what they're doing. life just does its thing.

you found the mark.