GenAI for Quantum Code Generation

Developing domain-specific datasets and evaluation pipelines for large language models focused on quantum code generation. This work studies whether curated quantum programming data can improve reliability, framework awareness, and hardware-aware code generation.

Current efforts include benchmarking domain-adapted models against state-of-the-art LLMs for PennyLane-centered and broader quantum software generation tasks.

Noise-Resilient Quantum Learning

Designing quantum learning models and training strategies that remain robust under realistic noise and imperfect quantum operations. This includes studying whether noise can sometimes aid optimization, improve generalization, or mitigate barren plateau effects in near-term quantum systems.

Current work explores noise-aware training, hardware-aware QML, quantum error mitigation, and robust hybrid architectures for deployment on NISQ-era processors.

Quantum Optimization and Applications

Exploring quantum machine learning and quantum optimization for real-world application domains including finance, healthcare, transportation, classification, and decision-making systems.

Research investigates how variational quantum algorithms and hybrid optimization workflows can provide scalable solutions under realistic computational and hardware constraints.

Quantum Resource Modeling and Benchmarking

Building unified classical–quantum cost representations for benchmarking hybrid quantum systems under realistic deployment conditions. This includes FLOPs-aware HQNN analysis, analytical runtime estimation, and hardware-calibrated evaluation pipelines for comparing quantum and classical resource trade-offs.

The goal is to establish realistic benchmarking methodologies that move beyond idealized quantum evaluations and enable practical comparison between classical and hybrid systems.

Resource-Aware Hybrid Quantum Neural Architecture Search

Designing multi-objective neural architecture search frameworks for hybrid quantum neural networks, jointly optimizing accuracy, trainability, parameter efficiency, and hardware cost.

This work explores cross-domain co-adaptation between classical and quantum components while incorporating realistic resource constraints such as FLOPs, qubit overhead, routing complexity, and execution latency.

Hardware-Aware Trainability and Expressibility of VQAs

This project investigates how hardware-aware compilation (transpilation) impacts the expressibility and trainability of variational quantum circuits. By benchmarking multiple ansatz families before and after transpilation. Preliminary analysis show that hardware constraints can significantly and non-monotonically alter these properties and that logical-level metrics are not reliable predictors of hardware-level behavior, emphasizing the need for hardware-aware design of quantum algorithms.

Quantum for Finance

Developing hybrid quantum-classical workflows for portfolio optimization, option pricing, risk analysis, and finance-oriented decision-making problems.

Current interests include quantum optimization algorithms, variational quantum circuits for financial modeling, and resource-aware quantum pipelines for practical financial applications.