UpGrade

Powering the next generation of adaptive experiments in education

🏗 EASI System Architecture

EASI connects researchers, adaptive algorithms, and learning platforms into a unified experimentation ecosystem. Click any node to explore each component — or scroll the cards below on mobile.

System Architecture Overview Click / tap a node · hover to highlight
UPGRADE INFRASTRUCTURE ADAPTIVE ALGORITHM 👩‍🔬 Researchers / Educators 🤖 MOOClet Engine Adaptive Algorithms ⚙️ Experiment Engine UpGrade (EASI) 🎯 Personalized Interventions 🔗 Learning Platforms 📘 OLI Torus CMU Platform 📐 MATHia Carnegie Learning 🚀 Your Platform Plug in here! 🔮 Future Partner 1 Coming soon 🔮 Future Partner 2 Coming soon ··· More Integrations 📊 Course Data 📈 Experiment Data & Analytics
📦

Node

Select a node to learn more.

Researchers
EASI Core
Adaptive Algorithm
Interventions
Platforms Hub
Partners
Your Platform / Analytics
Future
👩‍🔬

Researchers & Educators

Experiment designers

Domain experts who design adaptive experiments, define intervention logic, set research hypotheses, and interpret causal outcomes to improve educational effectiveness.

⚡ Adaptive Algorithm 🤖

MOOClet Engine

Adaptive Algorithms

A modular adaptive algorithm framework supporting contextual bandits, Thompson sampling, and other real-time personalization policies.

🏗 EASI Infrastructure ⚙️

Experiment Engine

UpGrade (EASI)

Core infrastructure that manages randomization, treatment assignment, and policy updates at scale across multiple learning platforms — without disrupting learners.

🎯

Personalized Interventions

Customized learning nudges, hints, scaffolding, and content variants delivered to individual learners in real time, driven by algorithmic decisions.

🔗

Learning Platforms Hub

Educational platforms that surface UpGrade-driven interventions to learners and stream back rich interaction data for continuous model improvement.

📘

OLI Torus

CMU Platform

📐

MATHia

Carnegie Learning

🚀

Your Platform

Plug in here!

🔮

Future Partners

Coming soon

📊

Course Data

Raw interaction data — page views, problem attempts, time-on-task, hints requested — collected across all partner platforms and funneled into the analytics pipeline.

📈

Experiment Data & Analytics

Aggregated, anonymized experimental results used to evaluate causal effects, update adaptive policies, produce publication-ready findings, and drive evidence-based educational decisions.


How It Works

LayerRole
Researchers / EducatorsDefine experiment designs, hypotheses, and intervention variants
UpGrade (EASI)Manages randomization, treatment assignment, and real-time policy updates
MOOClet EngineSupplies adaptive algorithms (bandits, contextual models) to UpGrade
Learning PlatformsDeliver interventions to learners and stream interaction data back
AnalyticsEnables causal inference, publication-ready results, and continuous improvement

Want to integrate your platform? UpGrade is open-source. Visit the UpGrade documentation to get started.