I'm building Learning Compass — a private, single-user system for turning scattered material into deliberate learning.

It is the main project I work on: one place to capture what matters, choose what deserves attention, learn at the original source, reflect, create structured notes and recall cards, maintain a personal knowledge map, and resurface useful ideas later.

capture → curate → consume externally → reflect → notes
        → review → map update → resurface

• Captures links, text, PDFs, videos, Telegram shares, and RSS entries into one library.
• Keeps commitment intentional with a small active queue instead of an endless read-later pile.
• Tracks learning sessions while consumption stays at the original source.
• Turns reflection into editable notes and recall-card drafts rather than generating them automatically.
• Uses learning history to maintain a personal knowledge map and improve what resurfaces.
• Creates source-faithful Arabic HTML and PDF reading companions when the original format gets in the way.
• Connects the whole workflow to Hermes for controlled agent automation.

Learning Compass is a PWA built with TypeScript, Preact, Hono, Cloudflare Workers, D1, and R2. D1 owns the application state, R2 stores large artifacts, and every client or agent uses the same validated Worker API.

Explore the Learning Compass repository →

I also build tools around the same interests: Learning & Visual Agent Skills, Eqraa, Kindle Remote, and Resurfaced.