Embedded systems researcher and builder
Self-powered TinyML · Custom PCBs · CPU/FPGA/Linux · Vision control
Email · Website · METAL Lab · GeekPie Association
Best Paper @ IEEE ICPS 2026 · Loongson Cup Second Prize (Captain) · TI Cup Shanghai First Prize (Captain)
I am a Computer Science undergraduate at ShanghaiTech University, an undergraduate research assistant at METAL Lab, and President of the GeekPie Association, where I lead embedded hardware initiatives. I work across circuit and PCB design, embedded firmware, computer architecture, FPGA/Linux systems, computer vision, and on-device inference.
-
PEAK Self-Powered TinyML Sensor Node · Sole developer and researcher
Built the complete system from piezoelectric energy harvesting and a custom threshold-gated power board to nRF54L15 firmware, INT8 inference, and BLE. The 23.7 KB model reaches 98.68% accuracy; the power board achieves 172 nA UVLO quiescent current. -
MIKU LA32R Out-of-Order SoC · Captain / CPU & Linux
Led a four-member team to the Loongson Cup Second Prize. Optimized the four-issue SpinalHDL CPU and built the FPGA/Linux system, reaching 100 MHz on Artix-7 with Linux 5.14 and Buildroot. CPU development · Linux kernel -
Battery-Free IoT Energy Management · Co-first author
Developed a supply-demand model, adaptive power path, and task scheduler for photovoltaic IoT devices, reducing system power consumption by 60%. The work received the Best Paper Award at IEEE ICPS 2026. -
TI Cup Ball-Balancing Vehicle · Captain / Vision & Mechanical
Developed the CanMV K230 vision pipeline and integrated the mechanical system, reaching approximately 1 mm static error at 88.9 FPS with STM32F407 closed-loop control. -
Anti-Poaching Risk Mapping · Sole developer and researcher
Building a spatiotemporal patrol-priority system with Prof. Fei Fang at Carnegie Mellon University and WWF, including spatial evaluation, rolling backtests, and 200 m to 2 km risk-grid generation. -
SparkLive · Team developer / System integration
Co-developed a local multimodal streaming system for NVIDIA DGX Spark using Qwen3.5, Whisper/VAD, FastAPI, React, and CUDA media processing. The project received an NVIDIA DGX Spark Hackathon Excellence Award. Digital-human workspace
- Hardware: Schematic design · Multi-layer PCB layout · SMT · Board bring-up · Power-path management · Signal conditioning · 3D CAD
- Embedded: C/C++ · nRF52/nRF54 · STM32 · ESP32 · K230 · Zephyr · BLE/Wi-Fi/MQTT
- Systems: Scala/SpinalHDL · Verilog · LA32R/RISC-V/x86 assembly · FPGA/Vivado · Linux/Buildroot
- TinyML & vision: TFLM · INT8 deployment · Feature extraction · CanMV · Python
BetterEAMS · Vivado on Parallels ARM64 · VEX Communication · Lab Management Copilot
Outside engineering, I enjoy Hatsune Miku, small robots, video production, and VOCALOID music production.
