How ritesh2006m/google works
Project type and purpose
• Score Lab is a full-stack, serverless web application built with Next.js (TypeScript) that automates grading of student answer-script PDFs using Google’s Document AI and the Gemini API. The system provides separate dashboards for admins, teachers, and students, stores data in Cloud SQL, and assets in a Google Cloud Storage bucket.
File structure
• Top-level “app” directory:
• app/api/* → Next.js API Routes (serverless endpoints for auth, student workflows, teacher workflows, general OCR)
• app/dashboard/* → React pages/components for Admin, Teacher, Student UIs
• app/user/* → Login and Signup flows
• app/layout.tsx & app/page.tsx → Root layout and landing page
• “components” folder → Reusable UI components (charts, form inputs, sidebar)
• “lib” folder → Utility modules for DB (Cloud SQL via Prisma/pg), Google Storage, and helpers
• Configuration and scripts: next.config.mjs, tailwind.config.ts, tsconfig.json
README for additional insights
• Describes core flows:
1) Teacher uploads question PDF
2) Student uploads answer PDF
3) Document AI OCR
4) Gemini API scoring & feedback
5) Persist results in Cloud SQL & Cloud Storage
6) Dashboards retrieve and display metrics
• Tech Stack confirms Next.js serverless model, Cloud Run hosting, Cloud Functions for AI orchestration, Cloud SQL, GCS bucket
a. Main components
• Frontend (Next.js pages & components)
• API Layer (Next.js API Routes)
• Cloud Storage Bucket
• Document AI Service
• Gemini API
• Cloud Functions (optional orchestration)
• Cloud SQL Database
• External Auth (e.g., JWT or session store)
b. Relationships and interactions
1. User (Teacher/Student) → Frontend UI
2. Frontend → Next.js API Routes (REST endpoints)
3. “Upload” endpoints → lib/googleStorage.ts → Cloud Storage
4. app/api/general/ocr.ts → Document AI OCR
5. OCR output → Gemini API evaluation
6. Gemini result → Cloud Functions (if heavy compute) or direct API Route
7. Persist feedback & scores → Cloud SQL
8. Dashboards → API Routes → Cloud SQL (queries) → Frontend
c. Architectural patterns and design principles
• Serverless micro-services using Next.js API Routes
• Separation of concerns: UI vs. data-access (lib/db.ts) vs. storage service vs. AI service
• Stateless frontends, stateful database
• Reusable UI component library for consistent design
d. Key technologies/frameworks
• Next.js (React + serverless API routes)
• Tailwind CSS for styling
• Cloud SQL (PostgreSQL)
• Google Cloud Storage bucket
• Google Document AI (OCR)
• Gemini API (scoring & feedback)
• Cloud Functions (for heavy AI orchestrations)
• Recharts for data visualization