Skip to content

Preet37/SAGE

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

81 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

SAGE — Smart Adaptive Guide for Educators

Closing the gap between educators and the people they teach.

SAGE is a platform built for teachers — not just students. It gives educators the tools to deeply understand their own subject matter, create interactive and factually grounded course material, and deliver it to students in a way that is engaging, personalized, and measurably effective.


The Problem

There is a persistent gap between what educators know and what students receive. Traditional tools force teachers to choose between quality and scale: a hand-crafted lecture reaches 30 students; a recorded video loses nuance; a textbook goes unread. None of these close the gap — they widen it.


What SAGE Does

SAGE gives every educator three core capabilities:

1. Learn Content — Deeply

Teachers use the same AI-powered tutor that their students do. Before teaching a topic, an educator can explore it through a Socratic AI conversation, run live simulations, view interactive visualizations, and verify their own understanding — all from a single platform.

  • Voice-based AI tutor (ElevenLabs conversational AI)
  • Interactive physics/math simulations with real-time parameter controls
  • Deep Dive mode — concept cards with linked papers, videos, and figures
  • On-device AI (Pocket Tutor) for private, zero-latency learning

2. Create Course Material — Publish It

Educators build courses directly inside SAGE. The AI assists with content generation, structure, and accuracy checking. Every lesson is interactive by default.

  • AI-assisted course creator with lesson structuring
  • Automatic groundedness scoring on every AI response
  • Multi-format document ingestion (PDF, DOCX, PPTX, MP4) via Cloudinary
  • Image enhancement, OCR, and AI classification for uploaded materials
  • QR-code shareable courses

3. Monitor Students — Stay Connected

Teachers see exactly where each student is, what they are struggling with, and how their cohort is progressing — without needing a separate LMS.

  • Real-time learner presence network (WebSocket peer map)
  • Per-student progress tracking across all lessons
  • Session history and chat logs per lesson
  • Skill assessments and quiz analytics
  • Shared courses with visibility controls

Platform Features

Feature Description
AI Tutor Groq-powered (Llama 3.3 70B) Socratic conversation with streaming responses
Voice Agent ElevenLabs real-time speech-to-speech; agent navigates the UI, opens graphs, adds notes on command
Interactive Simulations LLM-generated Plotly visualizations with animation, sliders, and real-time parameter control
Code Runner Multi-language execution (Python, C, C++, Java, Go, Rust, TypeScript, and more)
Deep Dive Rich concept pages linking ArXiv papers, YouTube lectures, GitHub repos
My Documents Cloudinary-powered upload, OCR, and semantic memory injection
Pocket Tutor WebGPU on-device AI — zero tokens leave the browser
Knowledge Wiki Curated interlinked knowledge base of 51 topics across 9 AI/ML subjects
Network Real-time peer presence map showing active learners
Multi-agent Fetch.ai uAgents for orchestrated, payment-enabled AI delegation

Hackathon Tracks

Track Implementation
Fetch.ai uAgents with Chat Protocol + Payment Protocol; multi-agent tutor delegation
Cognition LLM groundedness verifier on every tutor response; semantic memory (TF-IDF)
ZETIC WebGPU + @mlc-ai/web-llm Pocket Tutor — fully on-device inference
Cloudinary Document ingestion, AI enhancement, background removal, OCR for My Documents
Arista Networks WebSocket pub/sub learner presence network with real-time heartbeats

Tech Stack

Frontend: Next.js 16, React, Tailwind CSS, Plotly.js, @11labs/client, @mlc-ai/web-llm
Backend: FastAPI, SQLModel, SQLite, Groq API (OpenAI-compatible), Cloudinary SDK
AI: Llama 3.3 70B (tutor), Llama 3.1 8B (fast tasks), ElevenLabs (voice)
Infrastructure: Judge0 CE (code execution), Piston fallback, WebRTC (voice), WebSockets (presence)


Quick Start

# Backend
cd backend
uv sync
cp .env.example .env   # fill in GROQ_API_KEY, AUTH0_*, CLOUDINARY_*, ELEVENLABS_*
uv run uvicorn app.main:app --reload --port 8000

# Frontend
cd frontend
npm install
cp .env.local.example .env.local  # fill in NEXT_PUBLIC_* keys
npm run dev

App runs at http://localhost:3000.


Vision

"The best teachers are learners first."

SAGE exists to enrich educator quality by making deep, interactive learning accessible to the people who shape how others learn. When teachers understand their material at a greater depth — and can communicate it through tools that are visual, voice-driven, and factually grounded — the gap between educator and student closes.

Docs: Vision · Architecture · Course Creator

About

Resources

License

Code of conduct

Contributing

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors