
Onward is an AI-powered web application that helps internationally educated nurses prepare for Canadian job interviews. By combining research insights with AI-driven coaching, the platform improves confidence, communication skills, and cultural readiness.
The project was created for the BCIT Digital Design and Development showcase, themed around using AI to design solutions for underrepresented and disadvantaged communities.

The team explored how AI could support newcomers to Canada across areas such as services, housing, and employment. Employment quickly emerged as a critical milestone for financial stability and social integration. Within healthcare, where immigrants make up 25 percent of the workforce, internationally educated nurses were identified as a group facing significant barriers and in urgent need of support.
Immigrants make up nearly 25 percent of Canada’s nursing workforce. Despite their expertise, many internationally educated nurses struggle to secure jobs that match their qualifications.
In 2021, more than a quarter of recent immigrants were employed in roles below their skill level, underscoring systemic challenges in credential recognition, licensing, and adapting to Canadian workplace expectations.
Many IENs face long and costly licensing processes, often requiring extra exams, bridging programs, and courses. These hurdles delay re-entry into their profession and contribute to underemployment.
Canadian healthcare interviews emphasize soft skills, cultural competency, and local practices. IENs often struggle with unfamiliar formats, terminology, and ways of presenting experience.
Interview anxiety is heightened by unfamiliar hiring practices and limited culturally relevant preparation. Since the STAR method is often new to IENs, they struggle to communicate skills confidently without tailored feedback.
Programs such as the Foreign Credential Recognition Program address licensing challenges, but few initiatives focus on interview preparation and confidence building. This gap became the focus of our project.

More than one in four recent immigrants in Canada are employed below their skill level.
To better understand these challenges, surveys and interviews were conducted with both internationally educated and local nurses. Findings showed overlapping struggles with anxiety and preparation, but also revealed distinct barriers for IENs.
These insights highlighted the need for a platform that goes beyond generic interview prep by offering culturally aware, healthcare specific coaching.
I contributed as both a UX researcher and frontend developer.
Preparation should bridge cultural gaps, not just knowledge gaps.
Onward is an interview preparation platform designed to address the challenges faced by internationally educated nurses entering the Canadian workforce. Grounded in user research, the app combines AI-driven tools with culturally aware coaching to help users practice effectively and build confidence.
The platform centers on three core capabilities:
Two personas guided the design direction:
By aligning research insights with these personas, Onward delivers a tailored interview preparation experience that bridges cultural gaps, builds communication skills, and helps IENs present their experience with confidence.
Several AI-powered tools exist for interview practice, including Yoodli, Google Interview Warmup, and PrepMeUp. While effective for general job seekers, these platforms are not designed for healthcare or internationally educated professionals.
Key gaps identified:
Onward fills this gap by offering a culturally aware, healthcare specific interview preparation experience.
Our iterative design process combined research findings, user feedback, and usability testing. Three values shaped the solution:
Deliverables:
The Onward MVP was built with Next.js and React on the frontend, with Supabase handling authentication and file storage. My primary responsibility was developing the core functionality that powered tailored interview practice, including file uploads, real-time transcription, and AI-driven feedback.
The app was designed in modular components so each feature could be tested independently while contributing to the overall experience.
Users can upload resumes and job postings through a drag-and-drop interface powered by Uppy. Files are stored securely in Supabase Storage and passed into the AI pipeline.

The platform also records the user’s video responses using the device camera and stores the recording in Supabase Storage. Users can replay their interview session to review their non-verbal communication and overall performance.

Uploaded files are sent to RoughlyAI by passing their public URLs, where the service analyzes resume content and job posting requirements. It returns parsed JSON containing tailored interview questions.

After the practice session, the user’s transcribed responses are also sent to RoughlyAI, which compares them to the resume and job posting, returning structured feedback on content relevance, communication clarity, and alignment with job expectations.

Accessibility is not a checklist. It's what makes practice possible.
Wording guide user choices.
Testing showed that what felt obvious to us as builders was not clear to new users. Terms like practice interview and mock interview caused hesitation. Renaming them to practice mode and mock interview gave people a clearer sense of progression and made their choices more confident.
Design shifts in implementation.
Working as both researcher and developer meant translating designs into actual functionality. In code, small details such as how an upload flow handles errors surfaced gaps we had not considered in Figma. Experiencing these edge cases firsthand sharpened how I think about feasibility and designing with implementation in mind.
Clarity and context are accessibility
For internationally educated nurses, preparation was already high stress. Any added complexity such as unclear instructions or inconsistent labels raised the barrier further. This made accessibility feel less like a checklist and more like a matter of cultural context: clear, low cognitive load design was what made the product usable at all.
Together, these lessons also reminded me to stay realistic about scope, focusing on MVP features within a tight timeline so progress stayed centered on what mattered most to users.