Skip to content

EmoryHacks MLH Snowflake Winner: Syntra is an interactive knowledge graph that visualizes and guides your learning journey as a living forest. Map concepts, track mastery, and discover connections between topics through an intuitive force-directed graph interface.

Notifications You must be signed in to change notification settings

deviny411/syntra

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Syntra

AI-powered learning companion that transforms any subject into an interactive knowledge map. Visualize, connect, and master topics with personalized AI recommendations, mastery tracking, and seamless integration of real-world learning resources.


🎥 Demo

👉 Check out our demo here


📚 Inspiration

Learning complex topics often feels like navigating a maze: disconnected concepts, no clear path forward, and no feedback on what’s mastered. Syntra was inspired by the need for a tool that not only aggregates the world’s learning resources but actively helps users see and build their understanding—concept by concept, connection by connection.


🚀 What it does

  • Interactive Knowledge Graph – Visualize your knowledge as a dynamic, AI-structured map.
  • Automatic Connections – Gemini AI links new topics to related nodes and finds subtopics within each concept.
  • Resource Aggregation – Instantly pulls Wikipedia summaries, YouTube tutorials, and arXiv papers as you explore.
  • Mastery Tracking – Update your understanding with a single click—your progress is visualized and analyzed.
  • AI Recommendations – Snowflake logs every interaction; Mistral and Cortex AI analyze your patterns, recommending what to learn next and where to bridge knowledge gaps.
  • Subtopic Generation & Bridge Topics – Intra-Explore unveils granular subtopics, and AI builds bridges between seemingly distant areas—no more blind spots in your learning.

🛠️ How we built it

Frontend

  • React + Vite for a fast, interactive UI
  • D3 and XYFlow for force-directed graph visualization

Backend

  • Node.js + Express / TypeScript APIs for graph/knowledge management and resource aggregation

AI & Data Layer

  • Gemini AI for semantic node linking, bridge topics, and subtopic generation
  • Python microservice (FastAPI/Flask) for analytics, mastery calculation, and integration with Snowflake
  • Snowflake as the data warehouse, logging user activity and learning paths at scale
  • Mistral & Cortex AI for mastery scoring, pattern recognition, and real-time topic recommendations

APIs

  • Real-time scraping/fetching from Wikipedia, YouTube Data API, arXiv
  • Future plans: Khan Academy and news integrations

DevOps

  • Modular microservice structure
  • Code-splitting and RESTful interfaces
  • Local/remote mode support

🧱 Challenges we ran into

  • AI Integration – Prompt engineering for Gemini to give robust, controlled chains/subtopics and minimize hallucinations.
  • Analytics Pipelining – Designing interaction log/mastery calculation flows that work smoothly across REST (Node) and analytics (Python/Snowflake).
  • Graph Visualization – Efficiently rendering large, growing graphs with force-directed layouts and real-time edge/node styling.
  • Domain Scaling – Creating subject-agnostic routines for node-linking, subtopic extraction, and mastery analytics.

✨ Accomplishments that we're proud of

  • Reliable multi-level mastery tracking, supporting both manual and data-driven updates.
  • Seamless AI-driven graph linking and subtopic/bridge node generation.
  • Real-time integration of learning resources—articles, videos, papers—without page reloads or blocking.
  • Robust logging and analysis pipeline for personalized mastery and recommendation.

🧠 What we learned

  • The value of decoupling slow content (arXiv) from instant knowledge (Wikipedia/YouTube/IQ).
  • How to tune AI prompts for "just right" knowledge chaining, avoiding overly generic or overconnected topics.
  • That mastery is best represented not as a static value but as a dynamic function of actual user behaviors.
  • How enterprise-grade analytics (Snowflake) can power personalized learning at any scale.

⏭️ What's next for Syntra

  • Integrations – Add Khan Academy, news article APIs, and other educational resources.
  • Collaborative Analytics – Use Cortex to analyze aggregate user paths and success, feeding those insights back into recommendations for all users.
  • Deeper Semantic Search – Vertex AI / embedding-based search for resources and graph linking.
  • Mobile App & Offline Support – For universal, accessible, and context-aware learning.
  • Public API and Plugins – Help third parties extend Syntra’s brain into their platforms.

About

EmoryHacks MLH Snowflake Winner: Syntra is an interactive knowledge graph that visualizes and guides your learning journey as a living forest. Map concepts, track mastery, and discover connections between topics through an intuitive force-directed graph interface.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published