Page no. — 01Nyma
Brand & Design System2025 – 2026
NYMA
Nyma is a social resale platform for designer, luxury, and vintage fashion — people comment, save, and follow around the pieces, not just buy them. I joined as the only designer after the marketplace and auction architecture were built, and gave the platform what it didn’t have yet: a brand with a reason, a visual system, and design rules codified into the front-end repo.
At a glance
- Brand rebuilt from νήμα — Greek for thread
- 30–40 key pages designed by hand in Figma
- Design rules codified into the front-end repo
17brand-manual pages
Project Summary
Nyma is a 2C social resale platform for designer, luxury, and vintage fashion. Users don’t just buy and sell — they comment, save, and follow around the pieces, so the product has to feel like a community formed around garments, not a transaction site. When I joined, the marketplace and auction architecture were built and a website was live; what was missing was the design layer. The visual foundation was weak, the UI logic needed cleanup, and the founders liked the name Nyma without being able to say what it meant.
I traced the name to νήμα — Greek for thread — and everything hung from that: moodboards and AI-assisted direction studies, a 17-page brand manual with role-based color and an archival typographic voice, thirty-odd key pages designed by hand in Figma — and finally the whole system written into code, so the rules survive me and the team can build mobile without a designer in the room.
Architecture in place. Meaning missing.
A working platform that didn’t know what it was
I joined after the hard plumbing was done: marketplace and auction architecture in place, a website already live. And Nyma was never meant to be a listing board — people comment, save, and follow around the items, so it has to feel like a community formed around fashion pieces. But the visual foundation was weak, the UI logic needed cleanup, and the brand rationale was underdeveloped. The founders liked the name; the name didn’t yet mean anything. My scope became the whole design layer: brand direction, visual system, UI logic — and, later, turning all of it into rules a two-engineer team can build from without me.
νήμα — the name already knew what the brand should be.
Finding the thread inside Nyma
Researching the word, I found Nyma could connect to νήμα — Greek for thread. That one link gave the name narrative and visual potential it never had: garments carrying stories across owners, continuity instead of pure transaction, resale as something curated and cultural rather than clearance. I built the moodboards around that idea — the Fates spinning and cutting thread, Greek textile patterns, archival fashion photography — and the direction stopped being a taste question.
/ˈni.ma/ — Greek: “thread; yarn”
Directions as conversation material
In parallel I used AI to widen the search. I fed it my competitive research, the secondhand platforms I had studied, visual references, and the brand-language drafts I was writing, and had it generate brand directions, homepage narratives, positioning, and copy versions. That let us compare quickly whether Nyma should lean editorial like a fashion magazine, archival like a curated archive, transactional like a typical marketplace, or fashion-forward with louder style. The outputs were never final answers — they were conversation material that helped a name that merely sounded good become a grounded direction around thread, continuity, and curation.
Reads like a catalogue raisonné — numbered, provenanced, quiet. The system recedes; the objects carry.
lot numbersconditionprovenance
closest to the thread — this is the structure that shipped
Structural, not expressive — a brand designed to be maintained.
Rules that protect the objects
The manual I wrote defines Nyma’s identity as structural rather than expressive: restrained, consistent, and secondary to the objects it carries. It reads as operational rules — what must remain constant, where variation is permitted, where expression is intentionally limited. When uncertainty arises, priority goes to clarity, reduction, and structural consistency. The line I kept coming back to: this brand is not meant to be reinterpreted — it is meant to be maintained.
Color that signals, type that archives
Color at Nyma is role-based, never decorative. Ceramic Black contains; the archival whites surface; Ceramic Yellow appears only as a material trace; Activation Blue is reserved exclusively for interactive states. Any use of color outside its defined role is misuse — the palette signals structure, state, and continuity instead of expressing identity. Murecho carries the words in thin and regular weights — bold is nearly absent from the system — an archival voice that is quiet, precise, deliberately non-performative, with mono reserved for system data, the way an archive labels its objects.
The restraint is also hospitality. Nyma serves wardrobes that have nothing to do with each other — a couture archive, a designer drop, a worn pair of Levi’s — and the system has to receive all of them without re-styling itself. That is the real reason the rules are this quiet: inclusivity was the design decision underneath the design decisions.
Activation Blue #0D5EAF
Interactive states exclusively — the bid action, nothing else.
“Any use of color outside its defined role is considered misuse.” — brand manual, 3.1
Thirty-odd pages by hand; AI where hands weren’t needed.
The pages AI couldn’t deliver
Once the direction settled, the work shifted into production — and AI’s role had to shrink. For the major feature pages it could suggest structure, but not the designer-level layout, hierarchy, and editorial pacing the brand needed. So I designed and iterated the thirty-to-forty key pages by hand in Figma: marketplace and auction surfaces, listing and seller flows, onboarding, profile, messaging — the spine of the product.
Minimal enough for the items, editorial enough for a voice
Imagery was a different problem: a fashion resale platform needs a lot of it to feel alive, and there was no content library yet. Beyond public-domain and licensed references, I used AI-generated assets for mood imagery and atmosphere that matched the site’s tone while keeping copyright risk low. The judgment underneath everything was the trade between style and inclusivity — premium enough for designer-fashion users, welcoming enough for vintage and broader resale buyers. Most calls in the system were made to hold styles that share nothing: the same card, grid, and type have to serve an Hermès collector and a vintage seller without re-styling. That is why the language landed on minimal but editorial: minimal so the items carry the page, editorial so the brand keeps a voice.
the collector gets ceremony — lot number, stamp, reserve — from the same quiet shell
The design system left Figma and moved into the repo.
Decisions written once, reused everywhere
Nyma is a small company; a heavy, formal design system was never the point. But the product kept growing, and I didn’t want to re-make the same design decisions every time a new surface appeared. So I started codifying the existing system directly into the front-end repo: layout rules, typography, spacing, component behavior, reusable patterns — design decisions as code the product can’t drift away from.
Draft A — the editorial argument
color · commerce-forward
argues with imagery — colorful lots, magazine grids, the platform as a shop window
Draft B — the structural argument
mono · system-forward
argues with structure — fees, formats, verification; the platform as an institution
Figma stopped being the only source of truth
That changed how I work. Instead of designing in Figma and handing off, I moved across Figma, Claude Design, and Claude Code as one loop: explore a direction with AI, generate the code, fix details back in Figma, re-export and let AI refine — or edit in Claude Design, or adjust the repo directly. Figma stayed for visual refinement, but the codified front-end system became where design rules are preserved, reused, and extended. The lesson I keep re-learning: AI only scales you when the framework is clear. Unclear brand, inconsistent system, vague prompt — AI just generates more noise. Framework first; then it compounds.
A system that keeps working after I leave.
Prototypes and a roadmap the team can build against
The reason the system matters now: the team is starting their mobile build, and my contract is wrapping up. So I used the same AI workflow to prepare their continuity — AI-generated early prototype directions for mobile as starting points, and a product roadmap built from competitor study plus a structured conversation with AI, output as HTML, brought into Figma, and merged with the design system into a working prototype the team can build against. The codified system isn’t for me; it’s the layer that preserves visual and interaction consistency after I’m gone. That is the real test of whether a design system is doing its job.
Most memorable moment
When a name the founders just liked became the thread everything hangs from
Nyma didn’t need a logo first; it needed a reason. The founders liked the sound of the name, but nobody could say what it meant. Researching it, I found νήμα — Greek for thread — and the whole brand was suddenly in one word: garments carrying stories from one owner into the next, resale as continuity instead of clearance. Every decision after that pulled the same thread — the Fates on the moodboards, the role-based palette, the archival voice, the condition reports, and finally the rules written into the repo. And because the thread has to hold every wardrobe — designer, luxury, vintage — the system stayed deliberately quiet enough for all of them to enter.
The honest part: I’m not a trained graphic designer, and Nyma was my first time owning visual identity at this scale. Stakeholder feedback was positive and the visual quality moved far, but a sharper eye would have made some calls sharper still — next time I bring critique in earlier.
And I would codify from month one. The system I built at the end to hand off mobile is the system that would have made every earlier month faster. Same lesson as Vicino, from the other side: the earlier the framework exists, the more AI compounds. The later it exists, the more you are cleaning up retroactively.
νήμα — the thread holds