Work Publications Li, B. , Ding, K, Dera, D. (2025). MD-SA2: optimizing Segment Anything 2 for multimodal, depth-aware brain tumor segmentation in sub-Saharan populations. J. Med. Imag. 12(2). [DOI]
Li, B. , Serrano, K., Mazzaro, M., Wu, M., Wang, W., & Zhu, M. (2023). Identification of Cyanobacteria for Harmful Algal Blooms Research Using the YOLO Framework. IEEE UEMCON. [DOI]
Projects
Evidential Learning for Image Reconstruction ML/DL, CV/Graphics Uncertainty quantification for grayscale-to-RGB reconstruction using evidential deep learning.
SAM2LoRA: Parameter-Efficient Fine-tuning for SAM2 ML/DL One of the earliest LoRA adaptations for the Segment Anything 2 image encoder.
Molecular Docking QAOA Quantum QAOA-based molecular docking configuration search, developed at Regeneron.
OCaml Raytracer CV/Graphics Multithreaded raytracer in OCaml with a GUI, supporting meshes, textures, and custom camera controls.
Autograd from Scratch ML/DL, Systems Vectorized autograd engine replicating PyTorch backward pass functionality.
LLM Steganography ML/DL, RL Engineering and detecting steganography arising from misspecified reward signals in LLMs.
Adaptive Cache ML/DL, Systems ML-based cache eviction policy that learns from key usage patterns.
ML/DL Expository Paper + Resources ML/DL Work-in-progress expository paper with illustrations covering ML/DL approaches.
NJAS Symposium Judging App Web Dev Streamlined review for ~80 judges and ~200 student projects with Google Sheets integration.
See more on my GitHub page!