Health. Tech. Build. Lifestyle.
London | Dubai.

MSci Pharmacology at King's. Drug-safety data, a system I built to run my week, and plenty of lifting, photos and coffee.

About me See my work

Black-and-white London street with stone buildings and a black cab — London

A plane window at dusk, the last orange light on the horizon — In between

Driving through Dubai at night, towers ahead and the dashboard glowing — Dubai

Now

Updated October 2026

This month.

Studying Year 2, MSci Pharmacology — Endocrinology, Drugs & Disease, Research Skills

Building Mission Control: an assistant that prepares my day, and Notch, which puts it in my MacBook notch

Wellness Running in the mornings, lifting in the evenings, and setting up liftKCL

Working Front of house at King’s Sport, New Malden · Subject Ambassador

Mission Control

Every productivity app I tried was built for someone else. So I built my own.

Mission Control is the desktop app I run my week on. My notes are the source of truth; an assistant prepares drafts and tasks, and nothing goes out until I approve it.

Belal at his desk at night in headphones, working across two monitors and a laptop

Example: lecture cards

My notes Plain text
Local server On my Mac
Dashboard Today, Projects
Assistant Drafts, never sends
Me Approve or not

Endocrinology note: this week’s lecture has no cards yet.

Outside work

The rest of the week.

Wellness · Training most evenings.

A gym floor with treadmills, racks and murals

Mirror selfie in a gym, phone covering the face

Coffee · Coffee, and the places around it.

An espresso and a pair of earphones on a grey table, seen from above

A closed laptop and an espresso on a wooden café table, seen from above

Food · Meal prep for the week.

A plate of rice, spiced mince and roasted peppers

Photography · Shot on iPhone 17 Pro.

Looking up through a glass atrium roof between dark steel-framed walls

A dark grey Lamborghini on a London street

What I've found

Projects on drug-safety data. Three results that surprised me.

46% A painkiller looked barely riskier for your kidneys than paracetamol. It was twice as risky — the obvious analysis missed 46% of the harm. Read the case 80→54% I gave up coverage on purpose. The extra matches were wrong codes, and a wrong code looks exactly like evidence. Read the case 0.71 The most advanced model finished last. Finding drug side effects in medical text, a zero-shot LLM scored 0.71; a specialist model won with 0.83. Read the case

Writing

10 Jul 2026 I turned my data project into a tool — and deploying it taught me more than building it Substack ↗ 2 Jul 2026 Most-reported isn't most-dangerous — what the drugs themselves reveal Substack ↗ 18 Jun 2026 What 1.2 million reports of "drug ineffective" taught me about reading safety data Substack ↗

Looking for a 2027/28 placement year in pharma.

Drug safety, real-world evidence, clinical data. Happy to talk, or to show you Mission Control running.

hello@belalzaky.uk