LearnPod

LearnPod is the #1 project in my portfolio because it is the system I actually use to learn. It solves my own problem as a self-directed learner: articles, reels, and videos get saved with the intent to learn and then rot in bookmarks. LearnPod captures those raw captures as atomic Pods (one idea each), groups related Pods into Modules (mini-courses), and surfaces them back through a spaced-repetition review queue, streaks, and daily XP goals so that saved content actually becomes durable knowledge.

The 'why' of the design is learning science first, app second. Every structural decision is grounded in principles I can defend: atomic Pods enforce one-idea-per-unit so retrieval practice has a clean target; Modules provide scaffolded sequencing and prerequisite awareness; the review queue leads with retrieval over exposure (the single thing most learning apps get wrong); interleaved review questions mix 70% new / 30% from completed same-domain Pods to engineer desirable difficulties; and SM-2 spaced repetition schedules reviews based on quiz performance so the forgetting curve is fought, not ignored. The gamification layer — streaks with freezes and repair, milestone tiers at 7/30/100/365 days, a daily XP goal with Casual/Regular/Intense commitment tiers, and first-attempt accuracy bonuses — is a deliberate Hook-Model implementation (trigger → action → variable reward → investment) that turns 'I should learn this' into a two-minute daily habit.

The content production pipeline is what makes the system sustainable. Pods are authored in Markdown with frontmatter inside an Obsidian vault (one piece of the pipeline, not the whole story), enriched with research, then pulled through a custom TypeScript sync script (`scripts/sync-learning-content.ts`) that validates, types, and emits them as build-time app data — no database, no CMS, plain text as the source of truth. A second script (`scripts/generate-quiz-questions.ts`) generates retrieval-practice assessments. The result is a content factory: capture → author → sync → ship, with the PWA installable on any device and offline-capable for review sessions on the subway.

LearnPod has been shaped by four VP-level expert reviews (Education, Engagement, Design, Marketing) that I commissioned and acted on. The education review validated the SM-2 scheduler, interleaved review, and review-over-exposure home priority as 'Duolingo-grade foundations' and pushed the roadmap toward measurable learning objectives, prerequisite graphs, and assessment quality. The engagement review graded the habit loop B+/A- and guided the move toward a state-aware single-next-best-action home. The design review validated the Fraunces + Outfit type system, warm amber palette, and motion language as a rare 'premium, editorial' learning aesthetic. The marketing review reframed the home as a brand front door and surfaced the content-engine tie-ins (Pod of the Day, milestone share cards, OG previews). This is a project that has been stress-tested by real expertise, and the architecture reflects it.