Keren Zhang

Applied ML researcher with a Ph.D.

background and hands-on experience in generative AI, computer vision, and scalable image-analysis systems. I have developed PyTorch-based segmentation pipelines, conditional generative models for synthetic image creation, weakly/self-supervised learning frameworks, and user-facing scientific software tools. My work focuses on turning ambiguous research problems into validated ML workflows, with experience spanning model development, dataset curation, evaluation, visualization, and peer-reviewed publication.

I am interested in applying multimodal GenAI and computer vision methods to build high-quality, customer-facing creative generation systems.

Postdoctoral Researcher at Georgia Institute of Technology (2025-05 – Present)

• Proposed a self-supervised learning framework for mitigating ambiguity and data scarcity in weakly-supervised segmentation
• Built a physics-informed foundation model for multi-task biomedical image analysis, supporting segmentation, detection, and object tracking workflows
• Mentored graduate researchers on experimental design, ML pipeline development, and reproducible research practices

Graduate Research Assistant, Hang Lu's Lab at Georgia Institute of Technology (2017-09 – 2025-05)

• Established a scalable PyTorch-based framework for 3D neuronal segmentation and high-throughput image analysis
• Designed a conditional GAN model for synthetic data augmentation with image generation, resulting in consistent IoU gains (≈10%) across multiple data availability scenarios, compared to prior state of the art
• Designed and implemented an interactive simulation tool for neuronal morphology generation and structural visualization
• Engineered microfluidic devices that enable high-throughput imaging for C.elegans' developmental study

Undergraduate Research Assistant, Brain McCloskey's Lab at University of California, Berkeley (2015-08 – 2017-05)

• Discovered the presence of intermediate ion species and change in surface structure during and after polarization
• Quantified the electrochemical behaviors of pure aluminum electrode and the rate of hydrogen gas evolution
• Postulated the unusual mechanism of the aluminum dissolution and corresponding formation of hydrogen gas

Ph.D in Chemical Engineering – Georgia Institute of Technology (2017-09 – 2025-12)

BS in Chemical Engineering – University of California, Berkeley (2013 – 2017-05)