US patent holder, building AI that ships

From research
to revenue.

It began with the question a curious 16-year-old asks a computer: how do you actually work? I opened the machine to find out. That curiosity quickly became code, which turned into my first paid software, sold at 18, planting a permanent belief that what I build must have tangible value in the real world. From there, the compounding began. Code connected to design; design to film and music; music to acoustics; acoustics to a hardware patent and a company; the company to data science; and data science into the world's largest marketing network, WPP. As an enterprise AI technical lead, I built platforms that scaled to handle complex global data, contributing to major enterprise deals worth $3B+ in annual media spend for brands like DoorDash, Nestlé, SC Johnson, and Harley-Davidson. Today, I operate at the intersection of core AI development and deep tech through my independent lab, Duviloper. Rather than building wrappers, the focus is on original, high-defensibility technology. This architecture drives eQCX, our proprietary spectroscopy data converter currently shipping in the Microsoft Store, as well as breakthrough models capable of identifying antibiotic resistance from a single Raman spectrum in minutes.

Invest or partner See eQCX live
Live
eQCX in the Microsoft Store
Minutes
To a resistance call, vs days of culture
68×
Faster than manual spectroscopy prep
$3B+
Annual media spend of won deals supported
1 + 3
Patents granted + pending
Ventures

What I am building

One Core AI Lab, a portfolio of products, one proven playbook. Duviloper runs a clinical flagship, a shipped platform product, and audit services on the same self-auditing core. Every venture follows the same rule: something real ships, and someone pays for it.

Live · Microsoft Store

eQCX

The lab's platform product: the AI engine for spectroscopy data. Scans a folder of raw files, detects 6 modalities across 15+ formats, and returns ML-ready CSVs with publication figures. 100% local processing, byte-identical output, free researcher tier to enterprise on-premise.

Try it: eqcx.duviloper.com
Flagship · Patent pending

Raman-based AMR prediction

Reads antibiotic resistance from a single bacterial Raman spectrum in minutes, instead of 1–2 days of culturing. A 3-way fusion model (linear + CNN + band features) with leakage, capacity, and confound audits behind every claim, and flagship dashboards for lab teams.

For: pharma and diagnostics · dashboards at dml.duviloper.com
The Lab · Core AI

Self-auditing AI

Most labs optimize accuracy; Duviloper engineers trust. Leakage hunts, capacity and oracle tests, anti-confound design, and honest NULLs when the evidence is absent. Four pillars reach the world: AI, signal processing, biotech, and platform design.

For: safety- and compliance-critical AI · NDA-friendly
Proven · 2014 – 2021

Amorphous Technologies

Founded and ran for 7+ years. Invented and commercialized the world's first tunable ribbon microphone, earning a US patent, and directed teams across acoustics, materials science, and machine learning.

Proof: hardware can be patented, built, and sold
Track record · Now

WPP Media · AI Task Force

Technical Lead, building the enterprise AI knowledge platform of the world's largest marketing network (MVP retrieved 3× faster), with AI demonstrations behind pitch wins including DoorDash, Nestlé, SC Johnson, and Harley-Davidson.

Proof: enterprise AI that survives real clients
Reach

Brands my work has reached

Through WPP Media and Choreograph engagements, my data models, platforms, and demonstrations have been in front of:

DoorDash Nestlé SC Johnson Harley-Davidson BMW LVMH Mars Adobe Volkswagen Capital One HSBC

Pitch work I contributed to won accounts including DoorDash, Nestlé, SC Johnson, and Harley-Davidson, together worth $3B+ in combined annual media spend.

Playbook

The repeatable founder playbook

Amorphous proved the sequence in hardware. Duviloper is running it in AI: a patent-pending flagship, a shipped product, and paying services. It is how the next venture gets built.

01

Find the costly problem

Labs losing hours to data plumbing. Scientists doing software work instead of research. Go where the waste is measurable.

02

Prototype and patent

A working demonstration on real data, fast, with IP filed early. The moat gets dug before the market sees the product.

03

Ship for real

Production engineering, security, cost control, and a store release. Not a demo. A product a stranger can install and pay for.

04

Commercialize

Tiers from free researcher to enterprise on-premise, plus custom deployment services. Revenue before roadmap promises.

About

A founder somewhere between art and science

I have spent 12 years building enterprise platforms, deep-tech ventures, and data products. I founded and ran Amorphous Technologies for 7+ years, inventing and commercializing the world's first tunable ribbon microphone. Today I am Technical Lead of WPP Media's AI Task Force, building the network's enterprise AI knowledge platform, and I co-lead Duviloper, a Core AI Lab whose flagship reads antimicrobial resistance from a single Raman spectrum in minutes and whose product eQCX ships in the Microsoft Store.

I started as a programmer at 16 and shipped my first paid software at 18. Then came a decade of art: branding studios, film production, music, and audio engineering across Tehran, Dubai, and Los Angeles, serving 50+ clients. That decade is why my products are engineered like systems and presented like stories, and it is what carried me back to code: acoustics, a patent, and eventually AI.

A fundamentally sound design requires no supervision; it carries its own momentum, letting the architecture lead itself.

GrantedUS patent, tunable ribbon microphone
PendingFlexible conductive materials
PendingAuthentication system
PendingSpectroscopy device calibration
Contact
Building the next venture, or bringing AI to your science? Either way, let's talk.

Investors, partners, labs, and companies that need this kind of thinking all reach me the same way: email. Tell me the problem and I will answer with numbers, not decks.

Invest or partner LinkedIn