Repeated interface decisions across products
Tarjimly had grown across onboarding, language selection, requests, matching, profiles, training and enterprise workflows. Many screens were different, but the actions, forms, states and request details were closely related.
I wanted the team to reuse those decisions while leaving room for the different needs of mobile, web and enterprise users.
My role in building and maintaining the system
My work covered the foundations, components and Tarjimly-specific patterns, along with their use across complete journeys and design-to-engineering handoff.
The library needed to improve how we designed and delivered screens. Its value depended on whether the parts worked together in the actual products.
Role selection before and after the system
Role selection is one example. The user still needed to make the same choice; the system gave that choice a more consistent hierarchy and visual treatment.
This was the kind of repeated decision I wanted to resolve once, then carry into other parts of the product.
Organising foundations and semantic roles
I organised the system into foundations, primitives, components and product patterns. Brand colours and semantic roles were separated so values could be applied consistently to backgrounds, borders, inputs, errors and focus states.
That structure gave us a shared starting point without requiring a new component for every feature.
Primary · Background · Border · Input · Destructive · Focus
Readable copy. Clear hierarchy.
Component anatomy, hierarchy and states
Buttons covered hierarchy, size, icons and states. Inputs shared an anatomy for labels, helper text and errors across text, passwords and selectors.
Language selection became a reusable product pattern. It appeared across several journeys, so its behaviour needed to stay familiar even when the surrounding task changed.
Shared anatomy, with clear states.
Applying patterns across complete mobile journeys
I applied the system across onboarding, service entry, language selection, request setup and Spark. Complete flows showed where the parts needed to work together.
Request topics and translator requirements supported matching. Spark added clarification, confidence and human review states while keeping the same product language.
Adapting shared patterns for web and Reach
Web translation, session tables and live interpretation needed different information densities. Reach also had management and operational actions that did not belong in the mobile request flow.
I kept shared foundations and used layouts appropriate to each task. The system needed to connect the products without forcing them into identical screens.
Consistency and engineering handoff
The system improved consistency across Tarjimly’s mobile and web products and made design-to-engineering handoff clearer.
It supported products serving 700K+ people and 1,300+ organisations. Those figures show the scale of the work; they do not measure growth caused by the design system.
Governance and maintenance lessons
Naming became less consistent in parts of the library. That showed me where governance needed more attention.
I would take ownership, contribution rules, naming, versioning and deprecation further, alongside adoption tracking and clearer design-to-code mapping. Adding components is only useful if the team can find, understand and maintain them.
