Open to opportunities

David Bian

@davidbian

Message

I build production-ready Python and machine learning pipelines, combining research rigor with practical engineering.

Zimbabwe

Message

What I'm looking for

I’m looking for a role that blends research and engineering—shipping reliable Python ML systems and tooling, collaborating in HPC/cloud environments, and strengthening pipelines with clear logging, validation, and reproducible evaluation.

I’m a Duke University student building research-grade software with a strong engineering focus. As a Biomedical Research Assistant, I engineered a production-ready Python CLI that automated the CT-to-XCAT virtual imaging pipeline, reducing manual processing by 60% and increasing batch throughput by 3×.

In parallel, I develop and evaluate generative models for real problems, including building a custom PyTorch diffusion pipeline using Hugging Face diffusers and running GPU experiments in a Slurm-managed HPC environment. I also teach core algorithms, and I bring that clarity into how I structure parameter validation, logging, and error handling across my pipelines.

Experience

Work history, roles, and key accomplishments

Biomedical Research Assistant

Duke Center for Virtual Imaging Trials

Jan 2024 - Jun 2026 (2 years 5 months)

Engineered a production-ready Python CLI to automate a CT-to-XCAT virtual imaging pipeline, reducing manual processing by 60% and increasing batch throughput by 3×. Added JSON-driven parameter validation, structured logging, MAT-file generation, filesystem path resolution, type-safe input parsing, and error handling to improve robustness and debugging.

Python CLI Development CT To XCAT Virtual Imaging Pipeline Structured Logging Systems MAT File Generation Error Handling

Algorithms Teaching Assistant

Duke University

Jan 2024 - May 2026 (2 years 4 months)

Provided weekly recitations for 20+ students on algorithms including sorting, searching, dynamic programming, parallel processing, graph algorithms, and greedy algorithms. Graded weekly assignments for 30+ students and provided feedback and comments to improve submissions.

Algorithms Sort Algorithms Search Dynamic Programming Data Structures Recitation Instruction

Generative Model Research Intern

Duke University

May 2024 - Jul 2024 (2 months)

Built a custom PyTorch diffusion pipeline using Hugging Face diffusers to generate class-conditional minority-class images from pretrained UNet2DModel and DDPMScheduler checkpoints. Ran GPU experiments on a Slurm-managed HPC environment and evaluated diffusion-based augmentation versus SMOTE and standard data augmentation across imbalance ratios.