am.

AMMAN → HONG KONG

Adnan
Manna.

Signal hunter. Microwatt maximalist.
Professional BS caller.

Skip to the projects
ADNAN MANNA / 2026SCROLL INTO THE EYE ↓NO VIDEO. JUST CODE.
02 / THERE’S SOMETHING IN THE NOISE↓ SCROLL OR DRAG THE EXPOSURES

ONE FRAME DOESN’T TELL YOU MUCH.

Look.
Then look
again.

A bit more starlight. A bit less noise.

1EXPOSURESMOSTLY NOISE

A visual analogy. Simulated exposures; independent noise falls as 1/√N.

ASTRASCAN / ISEF ’26 FINALIST

No dip.
Still a signal.

A frozen transformer, pretrained on 1.5 million MACHO light curves, pointed at Kepler stars it had never seen.

It still told planet hosts from non-hosts.

0.615ROC-AUC
< 0.001p-value
ISEF project
03 / HUSHIN DEVELOPMENT / THIS IS THE TARGET

HUSH / THE ENERGY BUDGET

A year on
a coin cell.

Building AI that can listen for a year without a battery change.

DAY001/ 365
LISTENING
77 µW→365 days
1 µW10 mW

Budget sketch: 2,430 J ÷ time. Ideal runtime; battery losses aren’t modelled.

How much can a machine understand
on 77 microwatts?

Hush is an edge AI project for always-on sensing. The aim is to keep the expensive computation asleep until the input needs a closer look.

The first use case I’m working towards is listening for changes in machine sounds. The constraint is the entire system’s average power, not just the model.

CEA-Leti’s result ↗
04 / OFF THE CLOCKCOUNT THE CARDS. CALL THE BLUFF.

A GAME OF BS

Math with
trust issues.

My favorite card game.
Go on. Call the bluff.

ON THE TABLE
9♣  9♥  9♦

“Two nines.”

Count the cards. Then read the face.

HAND 01 / TWO NINESSCROLL TO MOVE AROUND THE TABLE

THE PROJECTS

What I’m
working on.

Some research. Some businesses.
One app for my entire life.

ME, WITHOUT THE CHARACTERS

I build things that hear the signals everyone else throws away.

I grew up in Amman and I'm now at HKU, in BGLIS, a brand-new program that mixes AI, humanities and business. My research found planet hosts in starlight that wasn't supposed to carry them. My startup bet is AI that runs for a year on a coin cell.

Different fields, same obsession: there's always more information in the noise than people assume. I go get it.

A FEW THINGS ALONG THE WAY

2025
→ 2026

  1. 2026

    Regeneron ISEF Finalist

    Physics & Astronomy · representing Jordan

  2. 2026

    Emergent Ventures Fellow

    Mercatus Center

  3. 2026

    JISETF Winner

    Jordan’s national science fair

  4. 2026

    Tomorrow’s Leaders

    U.S. Department of State scholarship

  5. 2026

    EV unconference, Copenhagen

    Also designed the winning BGLIS cohort logo

  6. 2025

    NASA Space Apps

    9th nationwide · ahead of 35+ university teams

  7. 2025

    Jordan Young Scientists Finalist

    SIRAJ · assistive navigation

HAVE SOMETHING IN MIND?

AMMAN ↔ HONG KONG

I’m
listening.

A project, a question,
or just a conversation.

Let’s talk FIND ME ON LINKEDIN