Orkhan Mammadov — Product Designer (AI & Creative Tools)

Genie, A node canvas for creators who don’t think in graphs

Product Design
AI / ML Interfaces
Prototyping
Design Systems

Most node editors are organised around how the system works. Genie, my own product, is organised around what artists want to do. I led its product and design for two years.

Client: Genie, my own product
Year: 2024 – 2026
My role: Founder, Product definition, Interaction model, Prototypes
Disciplines: Product Design, AI / ML Interfaces, Prototyping, Design Systems
Live: genie.art

Genie canvas on a laptop: text, image and reference nodes wired into a video node that has produced four frames

The node settings panel: model family and variant, iterations, creativity, seed, then style, object and character slots

A ControlNet node with the reference image on one side and its depth map on the other, toggled by Original / Depth

Ask Genie: the agent greets, offers two starting workflows, and receives a brief with two reference images attached

Ask Genie asking four clarifying questions, then proposing a better and a cheaper node flow, each priced in credits

The agent assembling a train-video flow on the canvas as ghost nodes while listing both options and their cost per generation

The Community panel: workflows, model catalog, artist models, widgets, creative coding and 3D assets, created by Genie or published by users

Genie home on a laptop: latest projects, recent canvases with previews, drafts, and the credit balance

Settings, Wish Points: a balance of 9046 points, the level 40 progress bar with the next level bonus, four achievement badges with their point rewards, and a meme coins history table

Settings, Invites: referral links with their percentage, invite earnings, and the option to withdraw or add them to the balance

Settings, Billing: the in-app balance, a top-up button and the purchase history

Hermes Cloud, An admin panel for an always-on AI agent

Product Design
AI / ML Interfaces
Design Systems
Mobile

Case Study

Nous Portal is one account for Hermes Agent, the Inference API and Hermes Cloud. I designed the product, from usage and billing to a three-step agent deploy, and created the Nous design system.

Client: Nous Research
Year: 2026
My role: Product Designer, Product design, Design system
Nous visual language and initial Portal concept: Leigh Taylor
Disciplines: Product Design, AI / ML Interfaces, Design Systems, Mobile
Live: portal.nousresearch.com

Hermes Cloud on two phones: the Agents list with an online Hermes Cloud instance, and the Deploying screen stepping through compute, container, gateway and agent online

Usage on a laptop before a plan is active: an upgrade prompt to start using Hermes

Usage on a phone with no agents yet, offering Hermes Web, Hermes App and Hermes Cloud

Nous Portal Billing on a laptop: current balance, top-up and subscription credits, plan card and automatic top-up

Nous Portal Settings on a laptop: linked accounts and email preferences

Agents on a laptop: an online Hermes Cloud instance with its controls, and a draft agent that can be started or deleted

The Agents list on a phone: My Hermes Cloud, online and healthy, region IAD, size Small

Empty state on a phone: create your first always-on agent in the cloud

Create Hermes Cloud instance on a laptop: region, name, model, size and the daily price running or stopped

Deploying screen: allocating compute, provisioning the container, starting the gateway, agent online

Deployed screen with two ways to start: Telegram or chat on the web

API keys on a laptop: keys with usage and revoke controls, and an example chat completions request

Tokens pricing on a phone: models, embeddings and tool pricing

ComfyUI, Linear Mode and Workspace for ComfyUI

Product Design
AI / ML Interfaces
Prototyping

Case Study

I helped design Linear Mode, where people run a ComfyUI workflow from a chat or a simple panel of inputs, and Workspace, one place for projects, canvases, workflows, models and assets.

Client: ComfyUI
Year: 2025 – 2026
My role: Senior Product Designer, Product design, Prototyping
Disciplines: Product Design, AI / ML Interfaces, Prototyping

ComfyUI dashboard on a studio display: templates, recent projects and recent team activity

ComfyUI timeline view: completed generations of four images with their prompt and parameters as chips

ComfyUI node canvas on a studio display: the workflow behind the chat

ComfyUI sidebar close-up: Dashboard, Recents, Library Hub, Projects and Canvases

Workspace switcher: a personal workspace and a studio workspace with four roles and their access levels

Library switcher: team libraries, a personal library and the community hub, each with an item count

Under the GAN, An immersive archive of early AI art, from 2015 to 2026

Product Design
Web
AI / ML Interfaces
Prototyping

Case Study

A web museum for the Jediwolf collection of early AI art: 198 works by 61 artists, from 2015 to 2026. Ten ways to browse, with ownership history read live from Ethereum.

Client: Own product, for the Jediwolf collection
Year: 2026
My role: Product owner, Design
Curatorial direction and historical framework: Kate Vass
Collector: Jediwolf
Infrastructure and deployment: Ahmet Ertas, Novexing
Disciplines: Product Design, Web, AI / ML Interfaces, Prototyping
Live: underthegan.art

The Museum view: a first-person three.js gallery, Chapter 01 Awakening 2015 to 2016, works hung on both walls with reflections on the floor

The Latent Space view: the collection as a navigable t-SNE field of works with a cluster map panel showing era distribution

The Sphere view: works and artist portraits wrapped onto a rotating globe

The Network view: the collection wallet as an on-chain graph, works drawn with their own thumbnails

The Canvas view: an infinite pannable field of works

The Timeline view: works ordered along the decade with era labels

The Index view: works, artists and dates as a plain catalogue

The console boot sequence that opens Under the GAN: loading archive, 198 works, 62 artists, eras calibrating, an Enter button

The Ten Ways of Seeing chooser on the landing page: Latent Space, Museum, Index, List, Horizontal, Grid Gallery, Canvas, Timeline, Sphere, Network

A work page: the piece centred with its media, model and facets

A provenance story beat on a work page: Since mint, this token has passed through 2 addresses, every hand it changed is written to the chain

The On-Chain page: the collection wallet read live from Ethereum, with balance, collections, transactions, token transfers and curated works

An artist page: Anna Ridler, 2 works in the collection, British, Forming, 2019

Torque, Brand, website and product design for an onchain growth platform

Brand & Identity
Web
Design Systems
Motion
Product Design

Case Study

A new visual language, marketing site and product direction for Torque Labs, the Solana growth platform behind 1,400+ campaigns.

Client: Torque Labs
Year: 2026
My role: Design direction
CEO, Torque Labs: Sheldon Smickley
Original site, 2024 to 2025: George Maroun, Vedran Sisic
Disciplines: Brand & Identity, Web, Design Systems, Motion, Product Design
Live: torque.so

torque.so homepage: The Onchain Incentive Infrastructure headline beside a Three.js ASCII helicoid, with the 1,400+ campaigns badge and $6B+, 5x, $10M+ stats

The hero as designed in the redesign branch: The Onchain Growth Engine, with the ASCII helicoid and the 89M+ transactions indexed badge

Six Primitives. In Your Product: leaderboards, raffles, rebates and more, each card carrying a procedural Canvas drawing

Effective Campaigns. Real Results: Winners Arc, Dynamic Rebates and Power User Conversion cards

Onchain Growth Intelligence: a terminal session inside the page where the Torque agent turns a request into a campaign

Platform page, Data Infrastructure: We index the chain so your team does not have to, with natural language queries, sybil protection, ROI simulation and real-time analytics

Platform page, The Console: build, launch and measure from one console, with campaign builder and analytics panels

The footer reveal: Incentive infrastructure for onchain protocols on a full blue field with the ASCII mark

The illustration kit: a grid of animated line-work visuals such as AcceleratorPath, AirdropShower, BuilderCanvas and CampaignRadar

The ASCII studio: drop an image and render it as a character field with density and charset controls

Design system, Color: four color families with ten grades each, blue, gray and cyan rows

Design system, Typography: Instrument Sans for display, Geist for body, Geist Mono for data, with the type scale table

Design system, Motion: purposeful, subtle and swift, with duration tokens

Design system, Grid and Layout: an 8px base unit, a spacing scale and responsive margins

Raisonne, A catalogue system for on-chain art, now becoming an open-source tool

Product Design
Web
Prototyping
Design Systems

Case Study

The catalogue behind orkhan.art: 4,139 works, 21 on-chain collections and 49 exhibitions in one CMS. It’s now being prepared as an open-source tool.

Client: Own product, orkhan.art
Year: 2025 – 2026
My role: Product lead, Design direction, Prototyping
Disciplines: Product Design, Web, Prototyping, Design Systems
Live: orkhan.art

The orkhan.art catalogue index: a masonry grid of series, each tile carrying its own artwork count and collector count

The orkhan.art home page: the positioning statement, a view catalog button and the showreel panel opening below it

The catalogue in gallery view: three works shown at full height, browsed as images rather than as an index of titles

The catalogue in table view: title, year, medium, tags and type columns, with exhibitions, installations and collections in one list and a series expanded into its works

A single artwork page: the piece at full height, with chain, token standard, year and marketplace listed next to it rather than behind a link

The Muraqqa Neural Impressions series page: a grid of works from the series, each carrying its own data panel under the image

The CV page: solo exhibitions on the left and group exhibitions on the right, each with venue, city, year and curator

The press and media index: coverage laid out as images and entries, drawn from the same CMS as the works

The insights landing: approach and methodology text above four categories, guild, analytics, leaderboard and network

The import at rest: pasted wallets, Ethereum and Base selected, and the six passes waiting, from deployed contracts to enumerating works

The review step: confirmed and suggested explained in plain language, with a co-authored series flagged, left unticked and waiting for a person

The review list with one series expanded: deployer, owner and ownership log, each with the address it matched and a transaction link

Shared contracts resolved token by token on Foundation, SuperRare and OpenSea, and below them a contract on Base carrying the artist name, unticked

The score panel: 22 of 22 found, none missed, 38 extra not listed on orkhan.art, first series at 1.5 seconds, 17.7 seconds in total

A portfolio preview generated from the ticked series, works loaded straight from their token metadata into a grid

What should we make?

I design the interface, then prototype it in React, Three.js, TouchDesigner or ComfyUI, whichever is fastest. Your team can try the idea before anyone writes a spec.

More about me