The AI coding mentor that actually remembers you — session after session.
🌐 Live Demo: recallcode.netlify.app 🔧 Backend API: recallcode-backend.onrender.com 📦 GitHub: github.com/GraceyDugar/RecallCode
LeetCode has 3,000+ problems. But it has zero memory.
Every time you open LeetCode, it treats you like a stranger. It doesn't know you always forget base cases in recursion. It doesn't know you've solved 40 array problems but never touched graphs. It doesn't know you submitted the same wrong pattern three times last week.
RecallCode fixes exactly that.
| Feature | LeetCode | RecallCode |
|---|---|---|
| Remembers your past mistakes | ❌ | ✅ |
| Personalised challenges based on weak areas | ❌ | ✅ |
| AI mentor that knows your history | ❌ | ✅ |
| Gets smarter the more you use it | ❌ | ✅ |
| Reflects on your mistake patterns | ❌ | ✅ |
| Swipeable challenge cards | ❌ | ✅ |
| Works across sessions | ❌ | ✅ |
RecallCode is powered by Hindsight — a persistent memory system that gives the AI a long-term brain.
Every interaction goes through three memory operations:
- 🧠 retain() — Every solution you submit is stored. Your mistakes, your language, your patterns — all saved.
- 🔎 recall() — Before every AI response, your full history is fetched. The mentor already knows you before you say a word.
- 🪞 reflect() — On demand, Hindsight analyses everything across all your sessions and surfaces recurring patterns you didn't even notice yourself.
- Paste any code, any language
- Get instant feedback that references your past mistakes
- Chat with a mentor who already knows your history
- Supports Python, JavaScript, C++, Java, C
- AI generates a personalised problem targeting YOUR weakest concept
- Swipe right to attempt, swipe left to skip — Tinder-style
- Every challenge is different because every student is different
- One click reveals your recurring mistake patterns
- Powered by Hindsight's reflect() — not just your last session, but everything ever
- Visual breakdown of what's holding you back
| Layer | Technology |
|---|---|
| Frontend | React 19, Framer Motion, CSS Glassmorphism |
| Backend | FastAPI, Python 3.14 |
| Memory | Hindsight (vectorize.io) |
| LLM | Groq — qwen/qwen3-32b |
| Frontend Deploy | Netlify |
| Backend Deploy | Render |
cd backend
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txtCreate a .env file in /backend:
HINDSIGHT_API_KEY=your_key
HINDSIGHT_BANK_ID=recallcode-bank
GROQ_API_KEY=your_key
uvicorn main:app --reloadcd frontend
npm install
npm startThis project was built for the Hindsight Hackathon by Vectorize.
Hindsight gives AI agents persistent memory — the ability to remember, recall, and reflect across sessions. RecallCode demonstrates what becomes possible when an AI truly knows its user over time.
Gracey Dugar — github.com/GraceyDugar | LinkedIn
B.Tech CSE (Data Science), Dayananda Sagar University | Co-founder, La Gravitea Deaf Café | MineSafeAI