Kashcool
AI-powered educational platform with real-time video, RAG tutoring, and AR content.
Kashcool is an AI-driven educational platform delivering personalized learning through real-time video, retrieval-augmented tutoring, and AR-enhanced study material for students who need more than static lessons.
The backend is a Django REST stack tuned for scale — API response times cut ~70% through query optimization, caching, and infrastructure hardening on AWS (EC2, RDS, S3, Route 53). Mentorship of junior developers sat alongside architecture work.
RAG tutoring grounds answers in each student's curriculum and session history instead of generic chat. Live classes use low-latency video with recording and interactive overlays, while AR visualizations help STEM topics land on mobile devices.
Adaptive learning paths use session analytics to sequence what comes next. The result was measurable: about +40% student engagement after the RAG tutoring rollout, with a backend that stays maintainable under growth.
RAG tutoring
Retrieval-augmented answers grounded in the student's own curriculum and past sessions.
Real-time video
Low-latency live classes with recording, playback, and interactive overlays.
AR content visualization
3D and AR-rendered concepts for STEM topics rendered on mobile.
Personalized learning paths
Adaptive sequencing driven by session analytics and performance signals.
Scalable AWS backend
EC2, RDS, S3, and Route 53 with staged deploys and observability.