Track: Concierge Agents — Kaggle x Google AI Agents Capstone 2025
Preparing for coding interviews is difficult because students waste time choosing random problems, don't track their weaknesses, and practice inconsistently.
This project solves it.
The DSA Practice Recommendation Agent is a multi-agent system that automatically:
✔ Analyzes your LeetCode profile
✔ Detects weak topics
✔ Recommends problems
✔ Fetches editorials
✔ Builds a 7-day practice schedule
✔ Tracks your long-term progress
✔ Provides a full Android app interface
- Stats Agent – Extracts coding stats
- Weak Topic Agent – Detects low-performing areas
- Recommendation Agent – Chooses the best questions
- Editorial Agent – Fetches explanations
- Schedule Agent – Generates a weekly plan
- Memory Agent – Stores user progress
- Progress Tracker Agent – Updates solved history
User Profile → Stats Agent → Weak Topic Agent → Recommendation Agent → Editorial Agent → Schedule Agent → Memory → Frontend App
- Python
- FastAPI
- Uvicorn
- Railway Deployment
- Custom agents and memory tools
- Android
- Kotlin
- Jetpack Compose
- Retrofit
- Material3 UI
| Endpoint | Description |
|---|---|
| full_profile | Fetch full stats + topics |
| weak_topics | Weak topic analysis |
| recommendations | Personalized problem recommendations |
| schedule | 7-day plan |
| mark_solved | Update progress |
| learning_path | Roadmap preview |
cd backend
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn backend.api_server:app --reload - Open
frontend/dsa_agentin Android Studio - Sync Gradle
- Update
ApiClient.ktwith backend URL - Run on device/emulator
- Stores solved problems
- Tracks weak topics
- Maintains session-based state
- Uses lightweight JSON memory files
The backend is deployed on Railway and accessible through a public endpoint.
MIT License.