Implements a grid-search method to find BKT parameters using a shared c library in Python. The shared c library implements the grid search to speed up the nested for loops. Functions to estimate BKT parameters, predictions, collapsed & non-collapsed rmse, auc_roc, accuracy and implement student-stratified k-fold cross-validation.

Multi-agent adaptive SAT tutoring system — a hierarchical Planner→Orchestrator→Specialist agent stack with Bayesian Knowledge Tracing, prerequisite-aware topic routing, and LLM-generated, symbolically-verified questions (FastAPI + vanilla JS).

🧠 AI-powered cognitive learning OS with knowledge graph diagnosis, Bayesian mastery tracking, spaced repetition study planner, and rubric-aware grading — built for Indian competitive exams (NEET, JEE, CUET)

ExemplAI is an example-based learning intelligent tutor for programming

Adaptiver Stochastik-Klausur-Coach, der Bayes-Inferenz nutzt, um deine Schwächen zu finden.

AI-powered adaptive tutor using PPO reinforcement learning + DKVMN knowledge tracing to personalize learning paths in real time. Dynamically selects concepts per student's knowledge state. Built with React, Django, PyTorch & trained on 131M+ interactions.

Modeling Knowledge Progression in MOOCs. An extended BKT model that integrates lecture engagement and time-gap data to improve knowledge state estimation in online learning environments.

Adaptive Platform for Learning Java powered by a hybrid BKT-RL model, featuring personalized challenges, performance tracking, and mastery-based progression.

An AI-Powered Adaptive Learning Platform Using Generative AI and Bayesian Knowledge Tracing(BKT)

Intelligent Weakness Reasoning Model — Multi-Signal Bayesian Engine + Interactive Next.js Dashboard with D3 Knowledge Graph, Severity Matrix, and Study Simulator

A real-time edtech agent with screen OCR, pluggable LLMs , Bayesian Knowledge Tracing (BKT) for adaptive quizzes, spaced repetition, concept graphs, and collaborative classrooms. Built with FastAPI, React, and Supabase.

Mastery Measurement API and CLI (TypeScript + Python) that fits Bayesian Knowledge Tracing and Item Response Theory models to learner response logs, for AI tutoring agents and edtech products that need real psychometric mastery scoring instead of raw percent-correct. npm + PyPI.

Closed-loop adaptive reading tutor for early & struggling readers (dyslexia-aware), built on Google's ADK. Keeps a per-child mastery model, picks the next phonics skill from real reading evidence, generates a guaranteed-decodable book, listens to the read-aloud, attributes every miscue to its grapheme, and adapts the next book session over session.

Open-source, offline-capable adaptive learning desktop app — BKT + SM-2 spaced repetition, hands-on terminal labs, BYOK AI tutor. MIT.