Ali Faraahani
Portfolio
Real productTabdealDesign system · DesignOps · Governance

A design system for crypto trading

I led the design system for Tabdeal’s crypto product: foundations, financial data, trading, wallet, identity, and operational patterns. The work connected reusable UI with the team’s review and handoff process.

What changed?

−20%

average task estimate per person

ClickUp · average estimate per person · one full quarter.

0

component-related UI revision loops

Limited to UI discussions about components, not all revisions.

↓

handoff cost, reported qualitatively

Lower handoff cost; no monetary amount or percentage is claimed.

From foundations to product flows

Built around the work of a crypto exchange

These explanatory examples were reconstructed for this case. They are not internal files or product screenshots. Financial values are illustrative, not live.

System architecture
  1. Foundations

    • Color
    • Type
    • Space
  2. Semantic tokens

    • Gain
    • Loss
    • Pending
  3. Components

    • Market row
    • Amount input
    • Status
  4. Product flows

    • Trading
    • Transfers
    • Verification

BTC / USDT

Sample price and volume

67,420.50

Price candlesTrading volumeFilled candles rise, outlined candles fallIllustrative data

Position lifecycle

Review

Show the estimated total and keep the unit next to the amount.

Quantity
0.005 BTC
Total
≈ 337.10 USDT

Explore components through product flows

Balances, addresses and responses are illustrative. Use these examples to explore the interface behavior.

Order entry & risk

Try buying and selling against a sample balance.

BTC / USDT

Increase the quantity to explore insufficient balance.

Sample balance
1,000.00 USDT
Order value
337.10 USDT

What does leverage change?

Sample margin
100 USDT
Position exposure
100.00 USDT
Effect of a 1% price move
±1.00 USDT
Leverage increases gains and losses

This illustration covers position exposure. Fees, funding and liquidation rules are excluded.

Deposits & withdrawals

Network selection, address validation and fees in one flow.

USDT
Sending and receiving networks must match

Check both the network and the destination before a transfer.

Receiving address

0x•••• •••• •••• ••••

Redacted address preview. Unusable for fund transfers.

Awaiting network confirmation

The balance updates after the required network confirmations.

Sample deposit
100 USDT
Available sample balance
500 USDT

Orders & details

Orders submitted in the example above appear here.

Orders matching the selected filter
OrderSidePrice (USDT)Quantity (BTC)StatusDetails
DEMO-101Buy67420.500.005Open
DEMO-102Sell68100.000.002Filled
DEMO-103Buy67420.500.5Rejected
01

Foundations & tokens

A shared language from visual values to product meaning and component behavior.

  • Color & typography
  • Spacing & composition
  • Semantic tokens
02

Financial data

Price, quantity, currency, and change need clear units and precision; color alone cannot carry their meaning.

  • Precision & units
  • Gain & loss
  • Tables & market rows
03

Trading & order entry

Reusable structures for entering an order, reviewing its consequences, and communicating execution state.

  • Buy / sell
  • Price & amount inputs
  • Order summary & history
04

Leverage & risk

Direction, exposure, estimated values, and warnings stay distinct, including the gap between submission and execution.

  • Leverage selection
  • Risk disclosure
  • Opening position
05

Wallet & transfers

Balances and transfer flows need consistent language for assets, networks, fees, and pending transactions.

  • Asset balances
  • Deposit / withdrawal
  • Network & fee details
06

Identity & access

The interface explains the next required action, missing information, and access to product capabilities.

  • Registration & KYC
  • Field validation
  • Verification status
07

Operations & dense interfaces

Tables, filters, and detail views organize the information needed to investigate and act on a record.

  • Filters & search
  • Data tables
  • Record details
08

Feedback & interaction states

A reusable component includes what happens before, during, and after an action—not just its default appearance.

  • Loading & empty
  • Success & error
  • Focus & disabled

The problem & the evidence

Where the problem came from

Reviewing the product exposed repeated interface decisions and inconsistent ownership. The work addressed both the component library and how designers and engineers used it.

  • Interface audit
  • Design documentation
  • Design reviews
  • ClickUp estimates · one full quarter
How it was measured
ClickUp estimates were reviewed across one full quarter, using average task estimates per person. I report a 20% reduction on that basis. Component-related UI revision loops reached zero and handoff cost decreased; those two outcomes are reported from the team’s process rather than a published cost model. The exact task count and numerical baseline are not retained here.
A/B testing

A/B readiness was part of the handoff checklist. That does not establish that the design system itself was A/B tested.

Measurement context

Estimates are planned effort, not tracked execution time or end-user task time. Zero refers only to component-related UI revisions, not every product or engineering discussion. No revenue or conversion lift is attributed to the system.

Decisions & leading the team

Before: UI delivery was the main expectation

At Tabdeal, design had been treated mainly as a UI-output function. A component library alone would not change that expectation. The handoff also needed to make the product decision, implementation context, and evidence clear.

Decision: make reuse part of the handoff check

I introduced a checklist asking whether compositions were built from the design system, which events could be collected, what success meant, and whether the work had an A/B test. Urgent work still needed a ten-minute team review before handoff.

After: review covered the decision and its implementation

Reuse and measurement became explicit review questions alongside the screens. Average task estimates per person fell 20% over a quarter in ClickUp. Component-related UI revision loops reached zero and handoff cost fell. Product and technical decisions still needed discussion.

The flow & the design logic

How decisions reached the product

  1. System

    Foundations

    Color, type, spacing, direction

  2. System

    Semantic tokens

    Action, surface, text, feedback

  3. System

    Components

    Behavior, states, composition

  4. Outcome

    Product flows

    Onboarding, wallet, trading

How the team kept it usable

  1. Team

    Design review

    Weekly or urgent 10-minute review

  2. Team

    Handoff checks

    Events, experiment readiness, reuse

  3. System

    Implementation

    Shared component contracts

These diagrams reconstruct the process and design logic; they are not product screenshots.

My direct work

  • Audited cross-product repetition and ownership gaps.
  • Defined token taxonomy, component contracts, and migration states.
  • Established the review and governance loop with design and engineering.

What I learned

A design system is an operating model with a visual output—not a component inventory.

More on the decisions

Consistency could not live in memory

Across onboarding, wallet, trading, and operations, repeated values and states were being decided screen by screen. The audit showed that missing ownership, naming, composition rules, and migration status mattered as much as missing components.

Three layers, one language

Foundations fed semantic tokens; semantic tokens fed components and flow contracts. Product patterns carried explicit states and accessibility behavior, so teams reused decisions rather than merely reusing rectangles.

Adoption was part of the design

Migration happened incrementally inside squad work. Weekly critique and an urgent ten-minute review kept events, A/B readiness, states, and implementation context attached to handoff. Four migration states made debt visible instead of pretending the system was instantly complete.

Choose “Save as PDF” in the print dialog to save the current page.