Open to opportunities

Ammar Ahmed

@ammarahmed6

Message

I build trustworthy medical AI systems, from generative models to clinical model audits.

United States

Message

What I'm looking for

I'm looking to build and evaluate trustworthy AI systems, particularly medical and generative models, while contributing to scalable ML infrastructure, responsible AI research, and clinically meaningful deployment.

At St. Jude Children’s Research Hospital, I build and evaluate medical AI systems for safe clinical deployment. I generated approximately 200,000 synthetic images from five medical diffusion models, helped uncover patient-level memorization, and evaluated vision-language models for demographic encoding and fairness disparities.

I orchestrate parallel machine-learning workloads across 10+ A100 GPUs on LSF-managed HPC clusters and built distributed training and inference infrastructure for a 3D generative model across five A100 GPUs.

Previously, as a graduate researcher at Johns Hopkins University’s Radiology AI Lab through the Microsoft Health AI Collaboration, I built the Emergent Language Symbolic Autoencoder in PyTorch to model hierarchical brain networks under weak supervision. My work spans trustworthy AI, generative AI, model evaluation, and reproducible large-scale ML infrastructure.

Experience

Work history, roles, and key accomplishments

Current

Software Engineer (AI Research)

Current

St. Jude Children’s Research Hospital

Mar 2025 - Present (1 year 7 months)

Generated approximately 200k synthetic images from 5 open-source medical diffusion models and helped design a privacy audit, finding patient-level memorization in 1 model and leakage susceptibility across all 5. Evaluated 8 state-of-the-art open-source medical vision-language models, finding that they encoded patient demographics and that stronger demographic encoding was associated with greater f