AI for science, built by a scientist.
PhD chemist. Three years inside pharmaceutical and industrial labs, installing instruments and building QC facilities. Since 2024, building AI agents and spatial software for the people who work there.
Thessaloniki, GreeceBuilding Varlis · Alepou
AI for science in mixed reality
Ask for a lab procedure and an AI builds it, straight onto your headset. A balance test where the computer reads the instrument itself is recorded; an HPLC pump service is in progress. Nothing in the headset app is specific to either: the same client solved a Rubik’s cube.
Alepou
Ask for an app and an AI builds it, straight onto your headset. One client on the wearable; every experience built on your own computer.
Universal client for wearables
Project Parallax
Mixed reality that recognises real laboratory equipment and guides the procedure around it.
Spatial computing
Varlis VR Training
B2B VR training software for scientific instruments and laboratory procedures. The first enterprise proof of concept was built for a global instrument manufacturer.
B2B VR training
Savans
AI inside an editor’s existing Word and tracked-changes workflow, with a human reviewing every edit. Used in paid work.
Applied AI
Spareleg
A side project: small science puzzle games. Quark Forge, a particle-physics colour puzzle, is out on Android; Firefly Relay is next.
Side project · puzzle games
Spatial AI at the bench
A guided eccentricity test on a real analytical balance, run in a headset with Alepou. The computer reads the balance itself over its serial port, and Claude checks the operator’s gloves and the final verdict.
One continuous take with a 100 g weight, edited for pace. The balance failed: front right read 0.8 mg below the centre, against a limit set strict on purpose at 0.3 mg, three times the readability. USP <1251> would allow 50 mg at this load.
Agentic spatial AI
Ask a question out loud in the headset, and an AI agent queries enterprise life-science data and puts the answers in the room: a drug-pipeline chart, 3D molecules, a patent family. This was the first public demo of Alepou’s spatial mode.
One take in the headset, about six minutes, lightly edited for pace. An independent demonstration on a commercial provider’s public life-science tools, not affiliated with or endorsed by the provider.
Built by asking
One spoken request, and the headset reads a real, scrambled cube from six photos, then guides it to solved one move at a time. The headset app has no cube in it: the workflow, its panels and the 3D cube were made by asking an AI.
Edited for length from one continuous take on a Meta Quest 3. The cube model, its moves and the panels come from the workflow at run time; the headset app is the same one that ran the balance test.
Universal client for wearables
Ask for an app and an AI builds it, straight onto your headset. The headset runs one Alepou app that never changes; the AI, each app’s logic and the connections to real instruments run on your own computer.
Each film here was made that way, on the same headset app. Underneath is Alepou’s open-source workspace for long-running AI work, where every change is a task tied to a git commit, so the AI’s work stays inspectable.
A 9×9 Go game played with pieces you pick up, a Japanese tea room with a cat, and the Mushroom March puzzle diorama. An HPLC pump-service walkthrough is in progress. Nothing about any of them is built into the headset app.
Spatial computing
Mixed reality that recognises real laboratory equipment and guides the procedure around it.
B2B VR training
B2B VR training software for scientific instruments and laboratory procedures. The first enterprise proof of concept was built for a global instrument manufacturer.
Applied AI
AI inside an editor’s existing Word and tracked-changes workflow, with a human reviewing every edit. Used in paid work.
Side project · puzzle games
A side project: small science puzzle games. Quark Forge, a particle-physics colour puzzle, is out on Android; Firefly Relay is next.
Chemistry is where I learned to work: form a hypothesis, design the experiment, measure honestly. Industry then taught me how instruments, methods and laboratories behave when real people depend on them.
| Code | Entry | Detail |
|---|---|---|
| PhD | PhD Chemistry, University of Hull, 2014–2019 | Molecular theranostics, photodynamic therapy, biological and organic chemistry. |
| PUB | First-author paper, Molecular Pharmaceutics | Peer-reviewed journal article from the PhD. |
| ACS ×2 | Two American Chemical Society National Meeting talks | Presented doctoral research in San Diego and San Francisco. |
| PAT ×3 | Named inventor on three patent families | Sporopollenin microencapsulation: WO2022200762A1, WO2022180405A1, US20240139335A1. |
| EOF | Pharmacopoeial analytical method | Developed for a pharmaceutical manufacturer; accepted by Greece’s National Organization for Medicines. |
I trained as a chemist at Hull, then spent a year and a half turning university research into patented products at a spinout. In 2020 we moved to Greece, and for three years I worked inside other people’s laboratories: installing and qualifying instruments, designing two pharmaceutical facilities from empty rooms and training the people who would run them.
That is where I learned what scientists need from technology, and how rarely software is built around it. Since 2024 I have been building it myself: AI agents that can be trusted with real work, and spatial software that sits beside real instruments.
2024 – present
Founder & AI Product Builder
Thessaloniki, Greece / international
Sep 2021 – Nov 2024
Senior Applications & Support Specialist
Greece · universities, industrial laboratories, pharmaceutical manufacturing and QC
Sep 2019 – Feb 2021
Project Manager
United Kingdom · university spinout · sporopollenin microencapsulation
2014 – 2019
Doctoral researcher, PhD Chemistry
Hull, United Kingdom · before that, MChem Chemistry (2:1), 2009–2014
