Scorpio
Scorpio is a research-driven educational platform engineered to transform physics instruction. By integrating a novel 4-layer AI constraint architecture with a high-performance, space-themed interface, Scorpio bridges the gap between traditional Learning Management Systems (LMS) and the dynamic cognitive requirements of physics problem-solving.
🏗️ Technical Architecture
Scorpio’s architecture is built for concurrency, type safety, and real-time synchronization.
🛡️ The 4-Layer Constraint System
At the heart of Scorpio’s AI tutoring capabilities is a proprietary constraint architecture designed to ensure pedagogical validity and eliminate hallucinations. The system enforces strict adherence through four distinct layers:
| Layer | Constraint | Purpose |
|---|---|---|
| 01 | Domain | Restricts knowledge exclusively to physics principles. |
| 02 | Pedagogical | Enforces the Socratic method; direct answers are prohibited. |
| 03 | Notation | Mandates proper LaTeX formatting and SI unit adherence. |
| 04 | Composite | Synchronizes all layers for research-grade tutoring. |
🔄 Data & Synchronization
- Real-Time State: Powered by Cloud Firestore with optimistic UI updates for zero-latency interaction.
- Asset Management: Client-side processed submissions (PDFs/Images) stored via optimized base64 in Firebase Storage.
🚀 Core Capabilities
🎓 Research-Grade AI Tutoring
The Scorpio AI Tutor is a context-aware pedagogical agent, not a generic chatbot.
- Context Retention: Maintains deep awareness of specific assignments and problem states.
- Adaptive Guidance: Dynamically scales hint complexity based on student performance.
- Academic Integrity: Hardened against jailbreak attempts and direct-answer harvesting.
🔢 Advanced Mathematical Rendering
Precision is non-negotiable in physics. Our custom rendering engine includes:
- KaTeX Integration: Blazing fast client-side LaTeX rendering.
- Visual Math Builder: An intuitive UI for constructing complex equations without raw LaTeX knowledge.
🌌 Immersive User Experience
A “Space-Themed” aesthetic designed to reduce cognitive load and boost engagement.
- Parallax Depth: Multi-layered backgrounds powered by Framer Motion.
- Glassmorphism: Context-aware backdrop-blur interfaces using Radix UI primitives.
🔬 Research & Efficacy
Scorpio includes a dedicated research dashboard to monitor the performance of its AI architecture. System metrics track:
- Rule Adherence %: The frequency with which the AI successfully maintains Socratic constraints.
- Response Quality: Automated evaluation of pedagogical relevance and clarity.
- Token Efficiency: Optimization of prompt length versus output quality to reduce latency.
Data Insight: Experimental results indicate that the Full Constraint Stack significantly outperforms standard models in educational utility, maintaining a high quality score across varying difficulty levels.
🛠️ Technology Stack
| Layer | Technology | Purpose |
|---|---|---|
| Frontend | Next.js 15 | Server Components, Streaming, and Routing |
| Language | TypeScript 5.7 | Strict type safety and developer ergonomics |
| Styling | Tailwind CSS | Utility-first styling with perceptually uniform colors |
| UI Library | Shadcn UI | Accessible, headless component primitives |
| Motion | Framer Motion | Physics-based animations and gesture handling |
| Backend | Firebase | Auth, Firestore (NoSQL), Functions, Storage |
| AI Model | Gemini 2.5 Flash | Multimodal reasoning and constraint adherence |
| Math | KaTeX | Fast, accessible equation rendering |
📂 Project Structure
├── src/
│ ├── app/ # Next.js App Router (Student/Teacher Dashboards)
│ ├── components/ # Reusable UI (Shadcn + Custom Math Components)
│ ├── contexts/ # Global State (Auth, Space Effects)
│ ├── hooks/ # Custom React Hooks
│ ├── lib/ # Core Logic (Firebase Admin, Gemini AI, Utils)
│ └── proxy.ts # Edge routing & middleware logic
├── public/ # Static Assets (Models, Architecture PDFs)
├── functions/ # Firebase Cloud Functions (Background Processing)
└── firestore.rules # Database Security Layers