王海飞 — 简历

FPGA & Embedded Systems: Verilog, PS-PL integration, AMD–Xilinx Vivado, Vitis, High-level synthesis (HLS), PCB prototyping

Electronic & Mechanical Design: Altium Designer, LCEDA (嘉立创EDA), SOLIDWORKS, AutoCAD

Systems & Programming: C/C++, Python, Linux kernel customization, Linux device drivers, Multithreading (OpenMP), SIMD (ARM NEON)

Supervisor: Professor Weibo GAO (wbgao@nus.edu.sg)

• Built and aligned a high-brightness, high-purity single-photon source using large-scale time multiplexing; implemented nanosecond-scale FPGA feed-forward control for real-time optical switching. Ongoing work.

• Calibrated a spin–orbit hyper-entangled photon system and developed control software for basis selection, quantum-state reconstruction, and data analysis. Manuscript under review at Nature Photonics.

• Contributed as a co-author to the experimental implementation of optimal quantum overlapping tomography by developing experiment-control and quantum marginal-state reconstruction software; the resulting work was published in Phys. Rev. Lett. 135, 060801 (2025).

• Leading a follow-up study that extends optimal quantum overlapping tomography to high-dimensional systems; implemented high-dimensional quantum marginal-state reconstruction and developed experiment-control software for basis selection and data analysis. Planned submission to Phys. Rev. Lett.

• Developed automation software for confocal-microscope sample scanning and data acquisition.

• Designed PCBs for mounting and wire-bonding superconducting nanowire single-photon detectors (SNSPDs), along with the associated readout electronics.

• Designed and set up a optical pulse-picker for mode-locked lasers using FPGA and Sagnac interferometry.

• Customized and integrated Analog Devices Linux kernel components and device drivers into PYNQ images for ZedBoard and PYNQ-ZU boards; exposed FPGA and RF-hardware control through user-facing Python APIs.

• Designed and integrated FPGA digital-baseband modules with ADC/DAC mezzanine cards, delivering an end-to-end software-defined radio prototype demonstrated at AMD–Xilinx Shanghai.

• GitHub: fm-demod-pynq-ad9361; fm-demod-rtlsdr-pynqz2; zedboard-adi-pynq

• Implemented and deployed a LeNet-5 inference accelerator on a PYNQ-Z2 board using C++ and Vitis HLS, covering convolution/ReLU, pooling, fully connected layers, preprocessing, and classification.

• Developed FP32 and INT8 quantized variants; the INT8 design retained 96.88% accuracy (97.65% for FP32) and achieved 417 frames/s versus 25 frames/s in software (approximately 17× acceleration).

• GitHub: lenet-hls-pynq

• Led development of an SSD MobileNetV1 accelerator on an Intel Cyclone V FPGA SoC, applying model pruning and quantization for FPGA deployment.

• Optimized the PS-PL data path using double buffering, OpenMP, and ARM NEON; reduced inference latency from >1,000 ms to 210 ms (approximately 4.8× faster) and won second prize.

Supervisor: Professor Yuanbo ZHANG (zhyb@fudan.edu.cn)

• Developed a Python package integrating LabVIEW APIs with scanning tunneling microscope (STM) topographic-scan and bias-spectroscopy data to automate tip-calibration decisions.

• Implemented sample-edge and surface-contamination detection, tip-sharpness and impurity assessment, and automated conditioning actions including bias-voltage pulsing and controlled tip dipping to reshape the tip.

Supervisor: Professor Shengjun YUAN (s.yuan@whu.edu.cn)

• Implemented a Trotter–Suzuki time-dependent Schrödinger equation solver using high-level synthesis (HLS) on FPGA, reproducing published results and achieving >200× speedup over a CPU baseline. The project was later entered as a competition submission at the 2022 AMD-Xilinx Summer School and won the outstanding project award.

• GitHub: tdse-hls-pynq

• GPA: 4.67/5.00; Distinguished Scholar

• GPA: 3.97/4.00; Distinguished Graduate; Outstanding Bachelor's Thesis

• Thesis: Hardware-software co-design for an AD9361- and PYNQ-based software-defined radio.