Research & Projects

Integration framework connecting quantum-circuit-optimizer (C++) with QubitPulseOpt (Python) for end-to-end fidelity analysis. 371 circuits analyzed, 23.1% mean gate reduction, hardware-validated on IQM Garnet 20-qubit processor. 252 tests. Submitted for publication.

Production-quality C++ quantum circuit compiler with OpenQASM 3.0 parser, DAG-based IR, 4 optimization passes, and SABRE routing for multiple hardware topologies. 340 unit tests. Designed for cross-layer fidelity analysis with QubitPulseOpt.

Physics-informed neural networks solving 6D basket option PDEs, volatility surface calibration, and advanced stochastic models. 306 tests, jump-diffusion, Heston stochastic volatility, American options, barrier options.

Python framework for quantum gate optimization using GRAPE with realistic noise modeling. Achieved 99.14% fidelity (77× error reduction). 822 tests, IQM hardware integration. Submitted for publication.

GPU-accelerated quantum state vector simulator using CUDA. Achieves 4.1× speedup over CPU at 22 qubits, competitive with NVIDIA cuStateVec. Features RAII memory management, density matrix simulation, and Lindblad noise models. 156 tests across 9 test suites.

GPU-accelerated Variational Quantum Eigensolver achieving 117× speedup on quantum chemistry simulations through JIT compilation, GPU parallelization, and MPI scaling. 99.4% parallel efficiency across 4 H100 GPUs. arXiv preprint submitted.

Real-time operating system for quantum pulse optimization and hardware control. Working GRAPE optimizer (99.9% fidelity), Rust HAL server, and Python client. 1,321 tests across Python and Rust.

Custom LLVM 18 pass analyzing loop structures and identifying unroll opportunities using Scalar Evolution for trip count analysis. Attaches full/partial unroll metadata based on loop characteristics.

ML framework for cloud base height retrieval from NASA ER-2 observations. GBDT R²=0.744 (per-flight shuffled CV), outperforming best CNN (R²=0.617). Catastrophic domain shift documented across flight regimes. Two first-author papers pending NASA approval.

Autonomous UAV perception system for detect and pursuit of target drones. Leading team of 7 as PI and perception lead. Accepted for presentation at SPIE Defense + Commercial Sensing 2026.

Open-source framework for building production-grade research software with AI coding agents. 5 core principles for correctness-first scientific code, comprehensive style guides, and the OpenCode Context Manager tool for efficient AI-assisted development.