Nura

Inspiration

Most brain-training apps give players disconnected mini-games and a mystery score, while computer science courses often teach algorithms through theory before intuition.

We wanted to bridge that gap.

Nura was inspired by a simple question:

What if people could experience an algorithm before learning its name?

Instead of memorizing BFS, DFS, Dynamic Programming, or Union-Find from textbooks, players learn naturally by solving interactive puzzles. Once they finish, Nura reveals the underlying algorithm through visual animations, pseudocode, and optional AI explanations.

What it does

Nura transforms real computer-science algorithms into engaging cognitive puzzles.

Players solve challenges inspired by:

Breadth-First Search (BFS)
Depth-First Search (DFS)
A*
Dynamic Programming
Union-Find
Constraint Satisfaction
Matrix Transformations
Deterministic Simulations

Every puzzle is evaluated using the real algorithm instead of AI guesses.

After each solve, Nura:

Scores accuracy, efficiency, and speed
Updates a private nine-dimensional cognitive profile
Shows an animated algorithm trace
Explains the solution with readable pseudocode
Offers optional AI coaching to improve future performance

Everything works offline using deterministic puzzle generation and local SwiftData storage.

How we built it

Nura is built as a fully native iOS application using:

Swift
SwiftUI
SwiftData
MVVM Architecture
Protocol-Oriented Programming
Deterministic Puzzle Generation
Graph Algorithms
Dynamic Programming
Constraint Solving
NVIDIA NIM
GPT-5.6
Codex

Each puzzle follows the same architecture:

Generator → Gameplay → Solver → Score Engine → Cognitive Profile → Algorithm Review

Every puzzle has its own seeded generator, formal solver, scoring pipeline, replay support, and visualization system.

Challenges we ran into

One of the biggest challenges was making eight completely different algorithmic puzzles feel like one cohesive product.

We also wanted every puzzle to be validated by deterministic algorithms instead of AI-generated answers while keeping the gameplay intuitive and enjoyable.

Designing a cognitive score that encourages learning without making intelligence or medical claims required careful consideration.

Creating an offline-first architecture while supporting optional AI coaching was another significant engineering challenge.

Accomplishments that we're proud of

Built eight playable algorithm-driven puzzle worlds
Created a reusable puzzle architecture
Designed the Mind Atlas progression system
Developed a nine-signal cognitive profile
Implemented deterministic algorithm validation
Added animated algorithm explanations and pseudocode replay
Built an offline-first experience with optional AI coaching
Created a polished native SwiftUI interface supporting Light and Dark Mode

What we learned

We discovered that people understand algorithms much better after they've already experienced them.

Instead of teaching BFS or Dynamic Programming first, letting players solve the puzzle before revealing the algorithm creates a far more memorable learning experience.

We also learned that consistency matters more than adding more puzzles. Every game should share the same learning journey: guide, play, score, explanation, and replay.