End-to-end agentic quantum-ML pipeline: PennyLane GPU quantum-kernel anomaly detection + LangGraph multi-agent system + locally-hosted Qwen3-14B, benchmarked across network-intrusion, LHC physics, and fraud.

Trainable, scalable quantum embedding kernels with variational parameters

Quantum Kernel Machine Learning for Drug Design A rigorous, end-to-end Qiskit implementation of quantum kernel SVMs for predicting blood-brain barrier permeability (BBBP) — a core ADMET property in CNS drug discovery — with three controlled experiments that actually test whether the quantum part is doing anything useful.

Quantum Robustness Score (QRS): a normalised metric for evaluating noise resilience in Quantum Kernel SVMs on NISQ hardware. Tested across 5 datasets, 3 noise channels, and ~2,000 simulation runs.

Quantum kernel SVM for rapid Ebola strain triage (Bundibugyo · Zaire · Sudan · Non-HF). ZZFeatureMap encoding of 20 clinical features. Motivated by the 2026 DRC/Uganda Bundibugyo PHEIC. PennyLane · scikit-learn · imbalanced learning.

Quantum-informed IBD modeling using ESM protein embeddings and Qiskit QSVC/quantum kernels; CLI for ingest → train → report.