OpenTutor Framework

Transforming discarded smartphones into private, offline AI tutors for equitable education.

OpenTutor is a lightweight, modular, offline-first framework that converts unused Android devices (2019–2023 models with 4–8 GB RAM) into personalized Socratic learning systems. It operates entirely on-device using small, optimized language models—requiring no internet access and preserving full user privacy.

---

Table of Contents

• Executive Summary
• Problem Statement
• Solution
• Key Differentiators
• Architecture
• Core Features
• Intended Impact
• Implementation Approach
• Current Implementation Status
• Quick Start
• Project Status
• Contact / Collaboration

---

Executive Summary

Access to high-quality, personalized education remains uneven, particularly in low-resource and connectivity-limited environments. At the same time, millions of functional smartphones are discarded each year.

OpenTutor addresses both challenges simultaneously:

• Repurposes e-waste into educational infrastructure.
• Provides private, offline AI tutoring for those without web access.
• Enables educators to create and share content without centralized control.

---

Problem Statement

• Digital inequity: Many learners lack consistent internet access for modern AI tools.
• Privacy concerns: Cloud-based tutoring systems require sensitive student data.
• Rigid pedagogy: Most edtech platforms prioritize answers over reasoning.
• E-waste growth: Millions of usable devices are discarded annually.

---

Solution

OpenTutor transforms low-cost, secondhand Android devices into offline-first Socratic tutors that:

• Run entirely on-device (no cloud dependency).
• Guide students through reasoning instead of providing answers.
• Adapt to learner behavior over time.
• Support extensible, educator-created content via plugins.

---

Key Differentiators

• Offline-first architecture: Fully functional without internet access.
• Privacy by design: No external data transmission.
• Socratic tutoring model: Emphasizes critical thinking and inquiry.
• Device reuse model: Converts e-waste into educational tools.

---

Architecture

graph TD
    UI["UI Layer<br/>Chat + Voice + Quiz"] --> Engine["Tutor Engine<br/>Socratic + Context"]
    Engine --> Inference["Inference Core<br/>MediaPipe • MLC-LLM • llama.cpp"]
    Plugins["Plugins<br/>Subjects • Pedagogy • MeshSync • Rap Hero • Lesson Forge"] --> Engine
    Storage["Storage<br/>SQLite + Plugins"] --> Engine

Loading

---

Core Features

• Plugin System: Modular subject and pedagogy extensions (e.g., math, literacy).
• MeshSync: Bluetooth-based transfer of educational content (hints, sequences).
• Rap Hero: Creative expression through educational rap generation.
• Lesson Forge: Enables students to create and share their own lessons.
• Adaptive Learning: Tracks learner behavior to tailor responses dynamically.

---

Intended Impact

• Short-term: Provide accessible tutoring tools in low-connectivity environments.
• Medium-term: Build a distributed ecosystem of shared educational content.
• Long-term: Establish a global, open infrastructure for personalized learning.

---

---

Quick Start

Prerequisites

• Android Studio (or Gradle 8+).
• Compatible Android device (2019–2023 model, 4–8 GB RAM).
• USB debugging enabled.

Build & Test Sequence

1. Clone the repo: git clone https://github.com/NewmanB1/OpenTutorFramework.git
2. Enter the directory: cd OpenTutorFramework
3. Build the APK: ./gradlew assembleDebug
4. Install on device: adb install -r app/build/outputs/apk/debug/app-debug.apk

---

Project Status

Current Phase: Early Alpha

• Core Framework: Functional (Builds and installs APK)
• Plugin System: Alpha (Basic loading supported)
• Basic Math: Partial (manifest.json and prompts present)
• MeshSync: Prototype (Early Bluetooth sharing)
• Inference Core: Integrated (MediaPipe / llama.cpp ready)

---

Contact / Collaboration

We are actively seeking pilot partners (schools, libraries) and educators interested in plugin development.

OpenTutor is built for communities, not platforms.