Cofounder and CIO, Entropy Labs

I still write the code that's expensive to get wrong.

I run product and engineering at Entropy Labs, a team of about a dozen. Most of my time goes to backend and AI systems where a small mistake turns into a wrong balance, a leaked tenant, or an agent doing something costly. On the side, I fix things in the open-source tools I use every day.

Halftone portrait of Muhammad Hassan Raza
170
merged PRs in kitty, calibre, libtorrent and 10 other projects
2 h to 25 min
manual onboarding flow, after the automation work at Entropy
750 to 230 ms
reporting endpoint, from reshaping the query rather than adding hardware
15+
specialist agents running in Obelisk

Entropy Labs Ignite acceleration round winner · Financing approved under Accelerate Prosperity · Exhibited at GITEX Global 2025 · Exhibited at LEAP 2026

Selected work

All work
Obelisk, halftone illustration

AI marketing platform

Obelisk

AI marketing platform with 15+ specialist agents. I lead product and the backend around them.

Retail operations platform

Polaris

Retail operations software, in daily use by shops since 2024.

Creator campaign platform

October

Campaigns between brands, agencies and creators, from brief to report.

Clinical callback workflow

Anatomia

Nurse callbacks and case review, with patient data handled like it matters.

Patches in tools I use

All 170 PRs

When something I rely on breaks, I read the source and send the fix. A few of the ones I'd point to:

Writing

Everything
  1. I Shipped a Race Condition That Oversold Stock (And Other War Stories) Oversold stock, silent chains, and the locks that fixed them.
  2. Agent Middleware: What Kept Obelisk's Agents in Line What kept a many-agent system cheap, recoverable and under control.
  3. LLMs Can't Save Bad UX Most failed AI features were failed product decisions first.
  4. The Best Way to Learn a Codebase Is to Break Someone Else's What 170 merged PRs in other people's code taught me.

Got a system that has to hold up?

I take on a small amount of outside work: architecture and AI reviews, hands-on builds, and advisory for teams shipping something real. I also taught university labs for a few years and still help students when I can.