About Underlayer

Learn things deeply.

Mission

Underlayer exists to solve a problem: most technical education is shallow. Courses rush through topics, skip foundational understanding, and rely on re-reading instead of active learning.

Underlayer takes a different approach. Every concept is taught from first principles, verified against authoritative sources, and tested until it sticks. We believe that understanding — not just exposure — is the goal of education.

How It Works

Structured Learning

Each concept follows a consistent structure: Why it exists, how to model it, the real-world details, concrete examples, and connections to other concepts. This ensures you understand not just what, but why.

Active Recall

After every lesson, exercises test your understanding. This isn't optional — it's how learning happens. Struggle is expected and normal.

Spaced Repetition

The FSRS v4 algorithm schedules reviews at the exact moment you're about to forget. This means minimal review time for maximum retention.

Deep Verification

Every hex dump, byte offset, and struct layout is verified against real specifications (gABI, ELF spec) or actual tool output (readelf, objdump). No invented data.

The FSRS Algorithm

FSRS (Free Spaced Repetition Scheduler) is the algorithm behind Underlayer's review scheduling. Unlike older algorithms (SM-2, Anki's default), FSRS models human memory as a multi-component system:

Stability — How long until you forget (in days)
Difficulty — How inherently hard the concept is for you
Retrievability — Current probability of recall

After each review, FSRS updates these parameters and calculates the next optimal review time. The algorithm adapts to your individual memory patterns, not just population averages.

Technology

Underlayer is built from scratch using the Chemical programming language. Here's what powers it:

Chemical

The programming language used for all platform code. Designed for educational tools.

SQLite / Turso

Dual-backend database: SQLite for local development, Turso (HTTP) for production. Same code, same queries.

Static HTML

Course content compiles to static HTML/CSS/JS. Works offline, loads fast, no JavaScript framework required.

Open Source

Built in public, verified by the community. Every design decision is documented and justified.

Principles

Depth over volume. One excellent course beats 100 mediocre ones.

Verify everything. AI-generated content is presumed incorrect until verified against authoritative sources.

Learning is the product. Every feature must answer: "Does this help someone understand something deeply?"

No passive reading. Every concept must be followed by active use.

Normalize struggle. Show "This is supposed to be hard" not "You're failing."

Your data is yours. Export everything. Delete everything. No tracking, no ads, no selling data.

Open Source

Underlayer is open source software. You can view the source code, report issues, and contribute on GitHub.