Technical build / AI education

Systems write-up. Static prototype and backend scaffold, not a visual product case.

Student-U. An AI coach that studies the class, not just the prompt.

Student-U is an AI study coach prototype for turning class notes into explanations, quizzes, and personalized study plans. The product direction is class-aware studying: course context, uploaded notes, professor comments, chapters, weak topics, and practice history all shape the generated support.

AI
Frontend
Product
HTML
CSS
Gemini
Claude
Codex
Cursor
Copilot

Role: Product designer + developer
Stack: HTML, CSS, JavaScript, Express
Focus: AI learning workflow
Backend: Gemini + Firebase scaffold

Problem

Generic AI study help loses the class context.

Students do not only need a quiz generator. They need help grounded in the actual class they are taking: the syllabus, the teacher's emphasis, recent lecture notes, weak areas, and what they have already practiced.

The prototype models each course as a living study portfolio, then uses that context to generate better explanations, quizzes, and plans.

Product concept

Each class becomes its own study workspace.

Student-U is built around a simple product bet: studying gets better when the AI knows the course. A biology class, a calculus class, and a writing class should not all produce the same generic flashcards from a pasted paragraph.

The intended workflow is class-first. A student adds course context, uploads or records study material, tracks weak topics, and asks the product to generate explanations, quizzes, and study plans from that specific class history.

That makes the project a good portfolio piece because it combines product thinking with implementation judgment: the experience has to feel easy for a student, while the backend has to keep AI keys and storage decisions out of the browser.

Build choices

Prototype first, secrets behind the backend.

Offline demo flow

The frontend can be opened directly in HTML and CSS and still show the core learning experience without backend services.

Backend-only AI calls

Live generation goes through an Express API path so Gemini keys are not stored in the browser.

Course material model

The backend scaffold includes course material metadata so uploaded notes and class context can later persist through Firebase.

What I built

A bridge between clickable prototype and real service.

Frontend prototype

A static app structure for the student-facing flow, designed so the product can be reviewed without requiring live AI services.

Express API scaffold

Backend routes prepared for generation, sessions, courses, material metadata, signups, and usage enforcement.

Firebase planning

Deployment and Firestore scaffolding for moving from local metadata toward real persistence and storage rules.

Responsible AI boundary

The frontend calls an API endpoint for generation instead of exposing Gemini or Firebase secrets to client-side storage.

Portfolio takeaway

This has the clearest product story of the newer GitHub projects.

Student-U is easy to understand quickly: students have scattered notes and weak topics; the product organizes class context and turns it into study support.

It shows that I can design around the class context a student brings to the task, not only around a prompt box.

Repository

Open the prototype structure and backend scaffold.

The README explains the static frontend, Express API scaffold, Firebase rules, course materials, and AI generation boundary.

Open GitHub

Orbsurv

A startup product platform with backend, web, and operations work.

Next case All work