Context

Context is an AI-based code learning platform built for a US client developing their MVP. The concept was simple but technically ambitious: instead of giving every student the same curriculum, the platform would meet each learner where they are. Students answer a series of questions about their background and coding experience, and the platform uses GPT to generate a personalized learning plan built specifically for them.

The platform also needed a voice assistant to guide students through content and questions, making the experience more accessible and interactive than a standard text-based learning tool.

Cenciss handled the full engagement: UI/UX design, front-end development, back-end architecture, AI integration, and deployment.

The personalization engine was the core challenge. Generating a learning plan that felt genuinely tailored rather than templated required careful prompt engineering and a question flow designed to capture meaningful signal about each student's experience level, learning goals, and pace preferences. A poorly structured input would produce generic output regardless of what the GPT API was capable of.

The user interface had to make the onboarding process feel natural. Students needed to move through the question flow without friction, understand what was being asked of them, and trust that the platform was building something useful. A confusing or clinical interface would undermine confidence in the AI output before the plan was even generated.

Voice assistant integration added a layer of complexity beyond standard web development. The assistant had to communicate generated content clearly, handle different types of questions, and integrate smoothly into the flow without interrupting the student's experience.

Security was non-negotiable for a platform handling student accounts and learning data. Authentication had to be properly implemented with JWT from the start.

The platform was built end-to-end on the MERN stack: React.js on the front end, Node.js and Express.js on the back end, and MongoDB for data storage. This gave the team a consistent JavaScript environment across the full application and a flexible data model suited to the variable structure of personalized learning plans.

The onboarding question flow was designed in close collaboration with the client, structuring questions to capture the right inputs for GPT prompt construction. The GPT API integration was built on the back end, with prompt templates engineered to translate student inputs into specific, actionable learning plans rather than generic recommendations.

The voice assistant was integrated to read questions aloud and guide students through the learning plan content, using the Web Speech API to keep the feature native and performant without adding third-party dependencies.

JWT authentication was implemented for secure user registration, login, and session management, ensuring student data and learning plans were protected and tied correctly to individual accounts.