Home / Projects / Guidly: AI Homework Feedback System
Guidly: AI Homework Feedback System
Guidly: AI Homework Feedback System is a project by Akash Renuka Ashok (Full-Stack Engineer & Product Designer). A privacy-conscious, open-source educational feedback system designed to identify misconceptions and support formative learning rather than a binary grade.
- Role
- Full-Stack Engineer & Product Designer
- Category
- Education AI
- Stack
- Next.js, Ollama, LLaMA 3, Qwen 2.5, Tailwind CSS, SQLite
The problem
Students receive 'wrong' feedback without understanding underlying misconceptions, while teachers lack insight into conceptual gaps making it difficult to decide what to re-teach.
What Akash owned
Owned product strategy, UX design, full-stack implementation, AI integration, security architecture and privacy-first data modeling.
Highlights
- Built and open-sourced a constrained AI feedback tool prioritising predictability and safe generation
- Built server-side AI misconception detection with static fallbacks so the system never fails silently
- Created a teacher misconception-visibility dashboard with AI-generated teaching recommendations
- Applied conservative AI usage principles: server-side only processing, always includes fallbacks
Outcome
Released an open-source learning tool prioritising clarity over grading efficiency, designed to help students understand misconceptions through targeted feedback.
More work by Akash Renuka Ashok:
Last updated