Project: Cognitive Reasoning and Adaptive Learning System

Track

Reasoning Agents (Azure AI Foundry)

Project Name

Cognitive Reasoning and Adaptive Learning Engine

GitHub Username

@DINITH-MV

Repository URL

https://github.com/DINITH-MV/Cognitive-Reasoning-and-Adaptive-Learning-Engine

Project Description

AI-powered adaptive learning platform with 7 specialized agents, real-time reasoning visualization, and personalized cognitive profiling.

Live Demo: https://cognitive-reasoning-and-adaptive-le.vercel.app/

CRACLE transforms education through intelligent multi-agent AI architecture. Seven specialized agents—powered by Azure OpenAI's GPT-4o—work together to create truly personalized learning experiences that adapt to each user's cognitive style, learning pace, and goals.

🌟 Key Features

• 🧠 7 Specialized AI Agents: Memory, Planner, Content Generator, Simulation, Evaluator, Mentor, and Orchestrator working in concert
• 🎬 Live Reasoning Visualization: Watch AI agents think and collaborate in real-time via WebSocket
• 🎯 Adaptive Learning Paths: Dynamically generated and adjusted based on your performance and cognitive profile
• ✍️ On-Demand Content: AI generates custom lessons, quizzes, and exercises tailored to you
• 🎮 Interactive Simulations: Decision-making scenarios that adapt to your choices
• 👨‍🏫 Personal AI Mentor: Context-aware guidance and explanations
• 📊 Cognitive Profiling: Tracks how you learn, not just what you complete

Demo Video or Screenshots

Screenshots: https://github.com/DINITH-MV/Cognitive-Reasoning-and-Adaptive-Learning-Engine/tree/main/Screenshots

Primary Programming Language

Python

Key Technologies Used

• React JS
• Python
• PostgreSQL

Submission Type

Individual

Team Members

No response

Submission Requirements

• My project meets the track-specific challenge requirements
• My repository includes a comprehensive README.md with setup instructions
• My code does not contain hardcoded API keys or secrets
• I have included demo materials (video or screenshots)
• My project is my own work with proper attribution for any third-party code
• I agree to the Code of Conduct
• I have read and agree to the Disclaimer
• My submission does NOT contain any confidential, proprietary, or sensitive information
• I confirm I have the rights to submit this content and grant the necessary licenses

Quick Setup Summary

git clone https://github.com/your-org/Cognitive-Reasoning-and-Adaptive-Learning-Engine.git
cd Cognitive-Reasoning-and-Adaptive-Learning-Engine
cp .env.example .env

cd backend
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
uvicorn app.main:app --reload --port 8000

cd frontend
npm install
npm run dev

Technical Highlights

What I'm Most Proud Of:

1. Live Reasoning Visualization 🎬
   The WebSocket-based real-time agent reasoning display is unique. Most AI systems are black boxes - CRACLE shows users exactly how 7 agents think, collaborate, and make decisions in real-time. This transparency builds trust and turns AI from mysterious oracle into transparent collaborator.

2. Sophisticated Multi-Agent Orchestration 🧠
   Seven specialized agents (Memory, Planner, Content Generator, Simulation, Evaluator, Mentor, Orchestrator) working in concert through a coordinator pattern. Each agent has ONE job, does it well, and agents can evolve independently. The orchestrator emits reasoning steps during workflows - elegant separation of concerns.

3. Cognitive Profiling Beyond Scores 📊
   Tracking HOW users learn (learning speed, comprehension style, analytical thinking, pattern recognition) not just WHAT they complete. Two users learning Python get DIFFERENT lessons based on their cognitive profiles. This is true adaptive learning.

4. Production-Ready Architecture ⚡
   Full async/await from database to AI (50x throughput vs blocking), SQLAlchemy 2.0 async, connection pooling, structured logging, error handling, migrations, monitoring, Docker containerization. Not a demo - ready for real users.

Interesting Technical Decisions:
Orchestrator Pattern Over Agent Mesh
Instead of letting agents call each other (N² complexity, debugging nightmare), one orchestrator coordinates all workflows. Memory → Planner → Content Generator flows through a single control point. Clear, observable, testable.

WebSocket Over Polling
For live reasoning, WebSocket provides instant updates with 1 connection instead of 1000s of HTTP polls. Session-based broadcasting means User A and User B don't interfere even on the same page.

Async-First Everything
AsyncAzureOpenAI, AsyncSession, async connection pooling. Result: 500+ req/sec per core vs 10 req/sec with blocking. Required for WebSocket scalability.

Unified BaseAgent Class
All 7 agents inherit from one base with shared Azure client, logging, metrics, retry logic, error handling. Change Azure endpoint once → all agents updated. DRY maximized.

Memory Agent as System Brain
Dedicated agent tracks all interactions, detects patterns, provides rich context to others. Mentor references past struggles, Planner builds on previous work, Evaluator sees long-term trends.

he Innovation:
Most systems: Static courses → everyone gets the same content
CRACLE: AI generates content on-demand → adapts to YOUR cognitive style → shows you how it thinks

The "wow" moment: First time watching agents collaborate in real-time to create YOUR personalized learning path. Transparency + intelligence + adaptability = transformative learning experience.

Challenges & Learnings

There was so many challenges

Contact Information

https://www.linkedin.com/in/dinith-mv/

Country/Region

Australia

This issue is a project submission rather than a scoped contribution task. Start with the linked repository's README.md and setup instructions; no specific file, test, entry point, or acceptance criteria are provided, so the work required and what would count as done are unclear.