Sajid Miya Computer Science and Engineering
Project Detail

SynapseLearn (Conversational AI Tutor)

SynapseLearn is a local-first, multi-agent conversational AI tutor designed to deliver personalized and context-aware learning experiences. The system integrates hybrid Retrieval-Augmented Generation (RAG), adaptive memory, and collaborative AI reasoning to move beyond traditional chatbot-based education systems toward an intelligent tutoring platform.

Open Project Back to Projects
SynapseLearn (Conversational AI Tutor)

core capabilities

Hybrid RAG Pipeline combining vector search, filtering, and refinement for high-quality responses
Dynamic Query Routing (Direct Answer, Retrieval, Web Search)
Multi-Agent AI Council for structured reasoning and tutoring strategies
Adaptive Memory System supporting both temporary and long-term learning modes
Self-Correction Mechanism to validate and refine generated responses
Hybrid Model Execution (Local via Ollama + Cloud APIs)
Context-Aware Learning via browser-based content capture
Student Evaluation Engine for quizzes, flashcards, and performance tracking

technologies used

Backend FastAPI (Python)
AI Orchestration LangGraph, LangChain
Model Execution Ollama (Local LLMs) + Cloud APIs
Frontend React.js, TypeScript
Data & Memory PostgreSQL, SQLite, Vector Databases (FAISS / ChromaDB)
Visualization Chart.js / D3.js

focus areas

Personalized AI tutoring through adaptive memory and context awareness
Efficient alternative to Graph-based RAG systems using dynamic pipelines
Scalable and privacy-focused local-first architecture
Bridging Generative AI → Agentic AI systems

system design

1 / 6
Use Case Diagram
Use Case Diagram
Level 0 DFD
Level 0 DFD
Level 1 DFD
Level 1 DFD
Level 2 DFD
Level 2 DFD
Hybrid RAG Architecture
Hybrid RAG Architecture
ER Diagram
ER Diagram