Publications & Open Source

Essential reference for candidates and teams: curated questions and step‑by‑step solutions across ML theory, deep learning, systems, and practical engineering.

QuantumLLMInstruct (QLMMI) is a pioneering dataset with over 500,000 curated problem-solution pairs for quantum computing, the largest of its kind. Built from over 90 seed domains and LLM-generated subdomains, it enhances instruction fine-tuning to improve LLM performance in quantum physics challenges.

Using a four-stage methodology, QLMMI covers areas like synthetic Hamiltonians, QASM code generation, and Trotter-Suzuki decompositions. It employs advanced reasoning techniques like Chain-of-Thought and Task-Oriented Reasoning, validated by a zero-shot Judge LLM for quality and reliability.

metalQwen3 is a GPU‑accelerated LLM inference stack using CUDA, Vulkan, and Metal Shaders for high‑throughput, low‑latency serving with custom kernels and memory optimizations.

OpenAI‑compatible local inference on vLLM with LMCache, tuned for RTX 5090, plus developer tooling and scripts for fast iteration.

Production pipeline for converting PDFs and images to searchable text/JSON with parallel processing and robust retry/logging.

Efficient quantum circuit simulation and QML on Apple’s MLX, optimized for Apple Silicon with hybrid quantum‑classical workflows.

QonFusion presents a novel strategy for generating Gaussian random variables using non-parametric quantum circuits, offering an alternative to traditional pseudorandom number generators. The research integrates Quantum Random Number Generators (QRNGs) into classical diffusion models.

This work demonstrates applications in Stable Diffusion and Brownian Motion, bridging quantum computing with classical machine learning through PyTorch integration.

This educational research explores the use of the Yao.jl quantum computing framework, written in Julia, for teaching quantum entanglement concepts to graduate students. The work provides a comprehensive pedagogical approach to understanding quantum entanglement.

The framework demonstrates how modern programming languages can make complex quantum concepts more accessible to students and researchers.

This innovative research applies quantum computing principles to music processing by developing a note detection algorithm based on the Quantum Fourier Transform (QFT). The work demonstrates the potential of quantum algorithms in signal processing applications.

The algorithm showcases how quantum methods can be applied to real-world audio processing tasks, bridging quantum computing with practical multimedia applications.

This medical AI research presents a sophisticated ensemble approach for detecting pulmonary nodules in medical imaging, combining the strengths of 3D SE-ResNet18 and DPN68 deep learning architectures for early lung cancer detection.

The work demonstrates state-of-the-art performance in medical image analysis, contributing to improved diagnostic accuracy in clinical settings.