Savva SicevsAI Design EngineerComplex AI systems,readable UI.
Control layers for AI agents, dense dashboards + workflow tools — so operators can read, trust, and intervene.
TaoMarketCapHow I work
“Setup took two weeks — I almost gave up before any value.”
“One dashboard I can show my team on Monday.”
“Templates would save me from a blank canvas.”
Latest Portfolio
Selected Work
Taomarketcap
A calm, scannable market-cap dashboard and trading surface for Bittensor's 100+ subnets — dense, real-time on-chain data plus money movement, made readable.
Rizzy
An AI lead-generation agent: conversational product experience and UI.
Desearch
A decentralized search narrative spanning web product and API docs.
Celium
A GPU marketplace dashboard with matched pricing and rental flows.
Drippler
An AI wardrobe concept: branding plus product UI across web and mobile.
Datura AI
Brand identity and product UI for an AI lab building on Bittensor.
CloakQR
QR-code cloakroom check-in — guests scan to drop off and collect coats, no paper tickets.
Built for the moment
a human has to decide
Agent control surfaces, dense dashboard UX, dev handoff / QA.
Read the Taomarketcap case studyServices
AI control surfaces
Agent consoles, approval queues, and data-heavy dashboards where a human supervises a fast-moving system — what happened, why, and where to step in — readable at table density.
Operator console
Supervising 6 live agents
Active agents
6
2 awaiting review
Avg response
142ms
Within target
Approvals
38
Today · 4 escalated
Throughput · last 24h
1,204 eventsNeeds review
Readability audit
A structured pass over a live agent or dashboard that works but doesn't read: friction, trust gaps, and clarity failures — delivered as quick wins you can ship now plus a prioritized roadmap.
UI System
A scalable component system with tokens, variants and specs that map cleanly to code — built at the density AI products actually run at, so the team ships consistent UI without re-deciding the basics.
Components
Synced library · 24 primitives
Neutral ramp
Accent
Spacing
Design Sprint
A focused 1–2 week sprint to move a hard problem — a new agent surface, a dense workflow — from ambiguity to validated, build-ready direction with prototypes and stakeholder alignment.
Define success metrics
QueuedMap user flows
QueuedEmpty + error states
ActiveMap user flows
ActiveResearch synthesis
DoneStakeholder kickoff
DoneNatural-language finder
DoneDev Support + QA
Handoff, pairing with engineers, and production QA so the design ships correctly — pixels, states, and interactions verified in the real build, including the failure states AI products live in.
Search field
180 × 40 px
Primary button
120 × 44 px
Toggle
36 × 20 px
Status tag
64 × 22 px
Avatar
40 × 40 px
Card
240 × 88 px
Search field
180 × 40 px
Primary button
120 × 44 px
Toggle
36 × 20 px
Status tag
64 × 22 px
Avatar
40 × 40 px
Card
240 × 88 px
const Button = styled.button` padding: 10px 16px; border-radius: 8px; height: 44px; `
Fast Prototyping
Rapid, working prototypes built with AI in the loop — idea to a clickable, real product in days, not weeks, so we can validate direction fast.
Applied 3 fixes
MRR
$48.2k
Users
3,180
Churn
1.8%
I design for the moment it ships, not the moment it’s presented
Stack & Tools
Claude Code
AI pair designer + engineer
Design and ship real React/TypeScript with an AI in the loop — this very site was built this way.
Ready to build faster?
Everything that builds this site, in one place — 12 components, each one click from your project.