Mastery measured by code, not clicks.
An engineering-grade learning portal powered by concept prerequisite graphs, isolated Linux sandbox validation, and Bayesian Knowledge Tracing.
Evaluation
Isolated Sandboxes
Mastery Engine
Bayesian BKT
Integrity
Keystroke Replay
$ docker build -t learner-image .
> [1/3] FROM node:20-alpine AS builder
> [2/3] RUN npm ci --omit=dev
> [3/3] EXPOSE 3000
Built for technical rigor, not rote memorization.
LearnLoop replaces subjective grading and passive reading quizzes with an authoritative, evidence-based loop designed for software engineers and trainers.
Prerequisite DAG
Curriculum concepts are mathematically modeled as a directed acyclic graph. Cycles are strictly rejected, ensuring students never encounter concepts without prerequisite mastery.
Bayesian Knowledge Tracing
Calculates mastery probability from verifiable execution evidence. Slip and guess parameters ensure lucky guesses do not falsely trigger advancement.
Hidden Server Test Suites
Learners test their code against public samples on demand. Official mastery assessments execute against hidden server tests that never leak to the client.
Keystroke Replay & Viva
Monotonic event buffers log fine-grained editing cadence. Anomalous paste bursts flag submissions for viva oral defense and trainer review.
Immediate feedback during practice. Authoritative judgment on submit.
Practice Mode: Rapid iteration with sample inputs, live compiler errors, and interactive diagrams in a sandboxed runtime.
Submit Mode: Asynchronous BullMQ worker queue runs complete hidden unit and integration test assertions in an isolated container.
Mastery Routing: The next concept or remediation is mathematically scheduled based on individual concept mastery state.
Role-Segregated Workspaces
Learner Portal
Personalized mastery progress, concept prerequisite visualizations, and active coding assessments.
Trainer Console
Batch cohort analytics, curriculum authoring, hidden test bank management, and viva integrity reviews.