Real products. Real engineering.

The problem

Most learning tools assume a single track. People who study several subjects end up juggling separate tools for roadmaps, flashcards, and progress, none of which share a picture of what they actually know.

What we built

• One loop for every subject: follow a roadmap, prove a concept with a spaced-repetition quiz, then level up.
• An XP and rank system tied to concepts retained, not time spent logged in.
• An AI coach inside the product, built on DODGE.
• A waitlist and open beta so real learners shape the product.

What it shows

Designing a systems-heavy consumer product end to end, building progression logic that means something, and putting AI inside a product instead of beside it.

The problem

A gym has owners, trainers, and members who each need a different view of the same business. Most tools force one view on everyone or split the data across separate systems.

What we built

• Three dashboard levels reading from one shared data layer, so each role sees exactly what it needs.
• One login for everyone, with access scoped by role.
• Modules for attendance, billing, equipment, analytics, and member management.

What it shows

Designing permissions and a data model around real roles, and turning one dataset into several focused dashboards.

The problem

Product sites make visitors hunt for answers. An assistant helps, but only if it stays on topic and its API key never reaches the browser.

What we built

• A chat widget that loads from a single script and can be branded for the host site.
• A server-side proxy that keeps the model API key out of the browser.
• Message length limits and trimmed conversation history to keep usage bounded.
• Fallback across several models, so a rate-limited model does not end the conversation.

What it shows

AI integration with sensible guardrails, a small serverless backend, and an embeddable front end. It runs on this site and inside PolymathOS.

The problem

Investing is usually taught as theory, and teens rarely get to practise before real money is involved.

What we built

• A path of eight steps, from Money Basics to Building a Portfolio.
• Lessons on risk, reading a chart, and compound interest.
• A fake-money practice market, so no real cash is at risk.

What it shows

Turning a dry subject into a game loop, and building an interactive simulation that feels like the real thing.