Open Source brilliant.org clone, community driven problems and courses

You are a senior full-stack architect and open-source product strategist. Design an MVP for an open-source Brilliant.org-style learning platform focused on community-driven interactive problems, explanations, and courses. Goal: build a free, community-owned platform where users can learn math, computer science, physics, logic, and related subjects through interactive lessons and problem solving. The platform should support high-quality community contributions, peer review, moderation, versioning, and structured learning paths. Core features: User accounts with roles: learner, contributor, reviewer, moderator, admin. Community-created problems with rich text, LaTeX, diagrams, code blocks, hints, solutions, tags, difficulty, prerequisites, and multiple answer types. Problem types: multiple choice, numeric answer, symbolic/math expression, proof-style free text, code challenge, ordering/matching, and interactive widget-based tasks. Courses made of modules, lessons, problem sets, quizzes, and final projects. Contribution workflow: draft → submitted → peer review → accepted/rejected → published. Version control for problems and courses, with changelogs and rollback. Discussion threads per problem/course. Reputation system for contributors and reviewers. Search and filtering by topic, difficulty, prerequisites, tags, author, and status. Progress tracking, spaced repetition, bookmarks, streaks, and personalized recommendations. Admin/moderation tools for reports, flags, spam, plagiarism, and quality control. Public API for problems, courses, users, tags, progress, and reviews. Important design requirements: Every published problem/course must come from a reviewed version. Problems and courses must be forkable, remixable, and attributable. Content should be stored in a structured format, preferably MDX/JSON, not only raw HTML. Keep interactive widgets separate from trusted core content. Add strong permission checks and audit logs. Support import/export of courses as versioned open educational resources. Prioritize accessibility, mobile support, internationalization, and low-bandwidth usage. Avoid copying Brilliant.org branding, proprietary content, UI, or problem text. Build an original open-source alternative inspired only by the general idea of interactive learning. Suggested tech stack: Frontend: Next.js, React, TypeScript, Tailwind CSS, shadcn/ui. Rich editor: TipTap or ProseMirror with Markdown/MDX support. Math rendering: KaTeX or MathJax. Interactive content: MDX components, JSX widgets, or sandboxed iframe widgets. Backend: Django 5.2 LTS + Django REST Framework, or FastAPI if prioritizing async APIs. Database: PostgreSQL. Cache/queue: Redis. Background jobs: Celery or Dramatiq. Search: Meilisearch for MVP; OpenSearch/Elasticsearch later. Auth: Django auth + OAuth, or Auth.js if using a Next.js-heavy architecture. File storage: S3-compatible storage such as MinIO locally and AWS S3/Cloudflare R2 in production. Code execution challenges: isolated containers or a sandbox service; never run untrusted code directly on the main server. Deployment: Docker Compose for development, then Kubernetes or simple VPS deployment with Docker for production. CI/CD: GitHub Actions. Observability: Sentry, OpenTelemetry, Prometheus/Grafana. Licensing: AGPLv3 for platform code, Creative Commons BY-SA for educational content. Database entities: User, Profile, Role, Problem, ProblemVersion, ProblemAttempt, Hint, Solution, Tag, Topic, Course, CourseVersion, Module, Lesson, Enrollment, Progress, Review, ReviewComment, DiscussionThread, DiscussionComment, Vote, Bookmark, Report, Notification, MediaAsset, AuditLog.

Comprehensive technical design package: system architecture diagrams (described in Mermaid/text), content storage format specification (MDX/JSON schema for problems, hints, solutions, widgets), versioning and forking model, permission/RBAC matrix for all five roles, review workflow state machine, widget sandboxing security model, audit log design, accessibility/i18n/low-bandwidth strategy, and 15+ Architecture Decision Records covering stack choices, licensing (AGPLv3 + CC BY-SA), and trade-offs. Delivered as a structured docs/ directory ready for the open-source repo.

Complete Django 5.2 models for all 26 entities (User through AuditLog) with migrations, constraints, indexes, and soft-delete/versioning patterns; entity-relationship documentation; full OpenAPI 3.1 specification for the public API covering problems, courses, users, tags, progress, reviews, search, and import/export of versioned OER packages; permission annotations per endpoint; pagination, rate-limiting, and webhook design. Includes seed/fixture scripts and a JSON Schema suite validating the structured content format.

Working Django + DRF backend: auth with roles and OAuth, problem and course CRUD with full version control (changelogs, diffs, rollback, forking with attribution), contribution workflow engine (draft → submitted → peer review → accepted/published), seven answer-type validators (multiple choice, numeric with tolerance, symbolic via SymPy, free-text proof queue, code challenge dispatch to sandbox service, ordering/matching, widget result verification), discussion threads, voting, reputation calculation, bookmarks, spaced-repetition scheduler, Meilisearch indexing, Celery tasks, Redis caching, audit logging, moderation/report endpoints, and a comprehensive pytest suite with factories covering models, permissions, workflow transitions, and API contracts.

Next.js + TypeScript + Tailwind/shadcn application: TipTap-based MDX editor with LaTeX (KaTeX), code blocks, diagram embeds, and hint/solution authoring; problem player rendering all seven problem types with sandboxed iframe widgets; course player with modules, lessons, quizzes, and progress tracking; review dashboard with diff view and inline comments; search/filter UI; profile, reputation, streaks, and bookmarks pages; moderation console; fully responsive, WCAG-conscious markup, i18n scaffolding with extracted message catalogs, and component/integration tests with Vitest and Playwright specs.

Original (non-Brilliant-derived) launch content authored in the platform's MDX/JSON format: 120+ problems across math, CS, physics, and logic spanning all seven problem types with hints, worked solutions, tags, difficulty, and prerequisites; three complete starter courses (intro logic, discrete math foundations, algorithmic thinking) with modules, lessons, quizzes, and final projects; six example interactive widgets (number line, graph plotter, truth-table builder, sorting visualizer, circuit toggle, code-trace stepper); plus contributor handbook, review rubric, style guide, and CC BY-SA attribution templates.

Docker Compose development environment and production-ready deployment manifests (Kubernetes plus simple VPS variant); isolated code-execution sandbox service with resource limits, language runners for Python/JavaScript, and a hardened job protocol; GitHub Actions CI/CD pipelines (lint, test, build, security scan, deploy); Sentry/OpenTelemetry/Prometheus instrumentation config; backup and S3/MinIO storage setup; OER import/export CLI; and full release documentation: README, installation guide, API docs site, security policy, governance model, and CONTRIBUTING guide for the open-source community.