cool description here
small paragraph self glazing my history as an artist
can specify materials, methods, locations here
Watercolor on cotton paper · 2025
Watercolor on cotton paper · 2024
Watercolor on cotton paper · 2024
can specify materials, methods, locations here
Oil on canvas · 2025
Oil on canvas · 2024
Oil on board · 2023
some catchy about here
Digital, Procreate · 2025
Digital, Photoshop · 2025
Digital, Procreate · 2024
$ whoami
annabelle chen — systems & hardware enthusiast
$ cat mission.txt
sentence on contents: research projects, personal endeavors like games
personal bio as a programmer
Ongoing and previous projects, starting with most recent.
PI: Dr. Matteo Riondato
Pursuing a senior research thesis in systems-rooted parallelization of uniform sampling of binary matrices. Potential explorations in the Rectangle Loop and Curveball algorithms, and implementations of data structures.
Check out our research group →PIs: Dr. Ryan Kastner, Dr. Curt Schurgers
Explored the deployment of a machine learning pipeline on embedded-edge computing devices paired with underwater camera technology for real-time computation of fish length and species. Focused on creating a method of unmanned data collection for deployment on underwater autonomous vehicles.
Read a co-authored paper on FishSense Mobile →e4e.ucsd.edu
Introducing FishSense Scout
Blog post with more project detail!
PI: Dr. Lillian Pentecost
Conducted partnered research on a project focused on customized hardware optimizations for an MCMC matrix transformation procedure (the Rectangle Loop algorithm). Applied computer systems principles to evaluate performance improvements and scalability of hardware-based solutions through the utilization of SystemVerilog in Vivado and high-level synthesis in Vitis HLS to design, simulate, and implement efficient modules on a Zynq Arty S7 FPGA
PI: Dr. Lillian Pentecost
Benchmarked ML tasks on resource-constrained processors, including digit recognition (MNIST) and object classification (EfficientNet) on Raspberry Pi 4 with camera. Deployed MobileNet models trained through Edge Impulse on Arduino Nicla Vision, optimizing for minimal memory usage while maintaining accuracy.
things just for me!