Quantum ML Classifier Platform

Screenshot

A full-stack classifier platform with a plugin architecture that lets new datasets be added without modifying existing code. Supports 6+ model architectures per dataset — including CNNs, SVMs, and quantum kernel methods via Qiskit and PennyLane — with real-time training curves streamed over Server-Sent Events. The evaluation pipeline includes per-class accuracy breakdowns, knowledge distillation, ensemble methods, and ablation studies. Features a 40+ component custom UI kit with dark/light theming, a draw-to-predict canvas for MNIST, and a form-based predictor for Iris. Covered by 425 tests across model architectures, training loops, API routes, and persistence.