I lead product design of a CPaaS platform built from scratch to replace a third-party solution and now in production. The work is under NDA, so this page covers the scope and my method, not the product itself.
Businesses send SMS — one-time passwords (OTP), notifications, marketing — through messaging platforms. A platform accepts traffic from many clients and delivers it through vendors and mobile operators. Around that sits a lot of operational work: client management, routing, pricing and balances, vendor connections.
The company ran it on a costly third-party solution and decided to replace it with its own platform.
Clients are managed in the system as a structure of several levels, and settings can be configured at every level or inherited from the one above.
ApproachI worked out the structure and the rules of inheritance with analysts before designing screens, and described every state in a behaviour specification.
ResultA flexible multi-level client structure with configurable and inherited settings at every level.
SMS routing is technical logic, yet the operations team is the one who configures it and diagnoses delivery.
ApproachI translated the routing logic into configuration flows and checked complex cases with analysts in AI-generated interactive prototypes before development.
ResultConfiguration flows with country-based rules and fallback vendors, and real-time monitoring of vendor connections — statuses, authentication failures and event logs — that centralises delivery diagnostics for the operations team.
Billing combines cost and pricing plans, time-bound tariffs, payments, credit limits and balances. Every figure is money, so the interface has to be exact and easy to trust.
ApproachI applied the clarity and trust patterns that consumer checkouts rely on.
ResultThe billing domain designed end to end, from plans to live balances.
The event data report is large and complex: many parameters, long periods and different needs for viewing and for download.
ApproachI designed viewing and downloading as one flow, with parameters grouped by purpose, periods managed in one place and templates that save a setup for reuse.
ResultAn interface for viewing and downloading the event data report, with flexible parameters, period management and reusable templates.
An internal platform is a working tool: people spend hours in it, not minutes. I shaped the design system around density, so screens hold more data and still read calmly.
A table, a form or a status behaves the same in every area, so learning one part of the platform teaches the rest.
Each component is designed for empty, loading, error and restricted states from the start.
Columns, filters and search set up for the task at hand, with export.
Large changes arrive in one file, and nothing goes live until it is reviewed.
Long settings stay readable, with validation that explains what to fix.
The same signals for state and access everywhere, so nobody has to guess what they can change.
Every feature goes through the same steps.
with the operations team: how they work today and where it hurts
with the product owner: what goes into the release and what waits
with analysts: rules and edge cases agreed before any screens
flows, screens and every state: validation, permissions, edge cases
complex logic validated before development
through development and release

