Jemima Abu is a frontend developer and product engineer whose work explains how web accessibility and browser-native features can reduce the custom code teams need to build and maintain. Across her product work and technical writing, she shows what that choice means in practice: content that survives a failed script, controls that work from a keyboard, and coding agents guided toward semantic HTML.
From a GPA calculator to product interfaces
Jemima began coding in 2017 after taking a C# course in her second year of university. Her first application was a command-line GPA calculator; an interest in how applications looked and felt led her toward frontend development. Her early career included ecommerce design and development in Lagos, remote work on websites for Savile Row shops at Sumosoft, and client interface development and frontend architecture at TeleSoftas.
At We Make Websites, she built Shopify ecommerce stores before joining CAIS in March 2025 to develop accessible interfaces for an alternative-investment platform. Her professional credentials include Google Developer Expert, Microsoft MVP and Certified Professional in Accessibility Core Competencies.
Letting the browser do the work
Jemima’s preference for simpler implementations predates coding agents. After an earlier portfolio proved slow, she rebuilt subsequent versions with fewer JavaScript dependencies. In her 2021 portfolio reconstruction, she used semantic HTML and CSS for expandable content, filtering and a carousel, keeping information available when JavaScript was disabled. Even animation followed that requirement: content became eligible to fade in only after the script loaded, so a failed script would not leave it hidden.
Her later writing applies that approach to common interface decisions:
Native interaction behavior: In her guide to reducing JavaScript in React applications, Jemima replaces custom accordion state with details and summary, and uses dialog for modal behavior that would otherwise require custom focus and keyboard handling. She distinguishes a blocking modal from a lightweight popover because the two need different behavior. JavaScript still has a role when requirements exceed what HTML and CSS provide, such as remembering a theme across visits.
Accessibility in generated code: Jemima uses coding assistants in her daily work, but her investigation of AI-generated modals shows why visual plausibility is insufficient. One generated close control was a span that keyboard users could not reach. Adding accessibility to the prompt improved the markup, yet the agent still needed guidance toward a native dialog. Developers must recognize these failures to prevent familiar, inaccessible patterns from recurring in generated interfaces.
In The Cost of Accessibility in AI, Jemima connects those implementation choices to the instructions and codebase an agent inherits. An inaccessible codebase supplies patterns to repeat; a vague design prompt leaves requirements unstated. Recreating native semantics, keyboard interaction and browser behavior in a custom component adds implementation and testing work that choosing an appropriate HTML element can avoid.
Her small experiment ran the same Figma implementation prompt with and without an accessibility skill. The run with the skill cost 58% less and finished in about a third of the time, but Jemima explicitly cautions that this uncontrolled comparison cannot establish that accessible development is always cheaper. Her practical recommendation is to make accessibility part of business requirements, design and reusable components, then carry it through implementation and QA. Audits and explicit agent instructions improve the starting point; keyboard checks and human review remain necessary because automated checks leave gaps. ARIA can help where needed, but cannot substitute for correct native behavior.
Teaching people to ship
Jemima’s educational work also addresses the practical obstacles of starting a career. In early 2021, she offered pair-programming sessions to developers setting up portfolio websites. She helped six developers get their portfolios online, then turned the experience into a beginner’s guide to domains, hosting and deployment. Helping others did not, as she had hoped, keep her own redesign on schedule—but it supplied the questions her guide needed to answer.
She has also taught programming fundamentals through Pearls Africa and GirlCode Academy, and web development to children at UNILAG CyberCamp. Her ADPList mentoring covers frontend development, accessibility, technical writing, public speaking and career growth.
Jemima Abu connects a small Figma implementation experiment to a larger engineering cost: agents can spend time rebuilding browser behavior, copying inaccessible conventions and creating repairs that better requirements would avoid.
One uncontrolled Figma implementation experiment reported 58% lower cost and about a third of the completion time with an accessibility skill; it does not establish a general cost advantage.
