Project case study · 01
Modi is the AI learning assistant within Modupath, an educational platform project focused on helping students learn, explore and plan their next steps.
Inside the project
The goal is to make Modi part of a broader learning experience. That means considering not only how the AI responds, but also how students navigate the product, access learning features and make use of the information they receive.
This case study describes the project's intended design and its known technology choices. Actual capabilities should be checked against the current implementation rather than assumed from the product idea alone.
Design and engineering
01
Students can have difficulty organising their learning, understanding unfamiliar topics and connecting what they study with their future goals. Modi was designed as part of a broader attempt to make digital learning guidance more accessible.
02
Modi is the AI learning assistant within Modupath. The intended experience is conversational: students ask questions, seek explanations and receive guidance that supports their learning rather than simply replacing their own effort.
03
The product is organised around a wider learning platform rather than treating a chat window as the entire experience. Learning tools, student-facing features and a consistent interface help connect the assistant to the wider product.
04
The application sends a user's question to its AI service, which generates a response that the interface displays. The project uses Groq for language-model inference. The exact context, data retrieval and error-handling behaviour depend on the implemented application code and configuration.
05
The broader project uses Next.js and React for its web interface, Firebase for authentication and Firestore data storage, and Groq for AI responses. These technologies serve different roles: interface rendering, account access, persistence and language-model inference.
06
AI response time is an important engineering challenge. A useful next step is to measure request duration, separate model-generation time from other server work, inspect logs and improve loading feedback. AI-generated answers also need careful evaluation because they can be incorrect.
What comes next
Future work can focus on faster responses, clear error messages, reliable AI behaviour, accessibility, testing and feedback from students and educators. These are development priorities, not claims that every improvement has already been completed.
Build learning software that helps students understand their options, develop useful skills and take greater ownership of their learning.
Return to the portfolio ↗