Gayatri AI Platform

Course-Independent Socratic Tutoring with Absolute Data Sovereignty

System overview

Architecture & Core Mechanics

Gayatri is an open, curriculum-free adaptive AI tutoring platform designed for schools, universities, and coaching institutes. Any institution can upload their own textbooks, syllabus, and lecture notes. The platform turns them into an intelligent, personalized, and trackable tutoring experience. Inference runs 100% on-device via local quantized SLMs, guaranteeing that no student data ever leaves the local machine.

Engineering Highlights & Capabilities

✓
4 Dedicated Portals: Student Learning Desk, Teacher Copilot & Intervention Queue, Parent Oversight, and Multi-Tenant Institution Governance.
✓
16-Step Pedagogical Turn Lifecycle: Enforces graduated Socratic questioning, misconception diagnosis, and zero answer leakage.
✓
Scoped Air-Gapped RAG: Institutions upload PDF, DOCX, or JSON; strictly isolated per-course retrieval with zero data leakage.
✓
Deterministic Mastery Engine: Ebbinghaus-decay spaced repetition scheduler (1d → 3d → 7d → 14d → 30d) and multi-factor competency scoring.
✓
Zero Cloud Egress: Runs 100% locally on standard CPUs via llama-cpp-python using quantized Qwen2.5 models.