Wakarumade
Learn until it clicks.
Created on 7th November 2025
•
Wakarumade
Learn until it clicks.
The problem Wakarumade solves
The problem it solves
Parents and teachers struggle to help young learners understand math word problems.
Most kids don’t fail because of the arithmetic — they fail because they can’t decode the language structure of the story.
This leads to:
- Guessing instead of reasoning
- Adults repeatedly explaining the same patterns
- No clear data on where the child got stuck
Traditional worksheets and drill apps don’t expose the child’s reasoning path.
They show answers, not understanding.
How this app helps (Current Capabilities)
Until You Understand/WAKARUMADE breaks each word problem into small, guided steps so children understand how to think, not just what the answer is.
Current users can:
- Follow a step-by-step scaffold tailored for early elementary level
- See the problem decomposed logically, one decision at a time
- Receive immediate feedback that nudges them back on track
- Have new similar problems automatically generated based on what they struggled with
- Benefit from backend logs that analyze which step caused confusion, and receive new variants targeting the same reasoning gap
This makes word problems:
- Less intimidating
- More predictable
- More accessible to ESL families
- Easier for parents/teachers to support without re-explaining everything
Future Directions (Not yet implemented)
Planned enhancements:
- Highlight key linguistic triggers in problems (e.g., “altogether”, “left”, “in total”) to teach the link between language and operations
- A parent dashboard showing:
- what patterns the child has mastered
- where they repeatedly get stuck
- simple explanations of the child’s reasoning weaknesses
- Progress tracking for logged-in students
- Personalized practice sets generated automatically from each child’s error patterns
- A library of problem types categorized by cognitive skill (comparison, combine, separate, equal groups, etc.)
Challenges I ran into
Challenges I Ran Into
- Implementing voice input for kids was tricky — raw speech is messy, so I built a pipeline that cleans and structures it before sending it to the AI.
- Designing persistence without login (kid-friendly, zero friction) required balancing simplicity with data-structure needs.
- Structuring logs so the system can detect where kids struggle and use that data for future problem generation was harder than expected.
Tracks Applied (1)
Best Use of GitHub
GitHub
Technologies used
Discussion
Builders also viewed
See more projects on Devfolio
