Luis Torres, PhD
Medical Physicist · Scientific Solutions Engineer
Experience
Flywheel
July 2022 – Present
Senior Scientific Solutions Engineer · Scientific Solutions Engineer
- Fine-tuned MedSigLIP to classify MRI contrast weighting from pixel data, demonstrating limited-data adaptation of medical foundation models to real-world tasks.
- Developed a retrieval-augmented generation (RAG) chatbot proof of concept that uses internal documentation to support developers with pipeline design and troubleshooting.
- Designed and developed automated release-note generation using LLMs in CI/CD pipelines.
- Lead design and deployment of scalable, containerized imaging workflows across multiple modalities (MRI, CT, PET, ophthalmic).
- Primary technical partner to industry and academic sites, translating research algorithms and scientific requirements into automated, validated imaging workflows for multi-institution studies, ensuring reproducibility and data integrity across sites.
- Built modular workflows for image curation, quality control and quantitative analysis across clinical and preclinical datasets.
- Oversaw ingestion and curation of external imaging datasets, directing contractor teams to ensure data quality, harmonized metadata and reproducibility standards for downstream analysis.
- Developed and validated modality-specific classification approaches for CT and MR, applying imaging physics to improve labeling consistency, robustness and interpretation.
- Consulted for imaging scientists, providing hands-on guidance in rule-based classification for MR, CT and PET.
- Mentored scientific engineers on designing reproducible containerized workflows, development best practices and CI/CD testing strategies.
- Designed custom Gears for advanced medical image processing, including segmentation, registration and feature extraction.
- Authored best-practice documentation for workflow containerization.
University of Wisconsin–Madison
August 2016 – June 2022
Graduate Student Researcher
Completed a PhD in Medical Physics focused on quantitative MRI biomarkers and advanced reconstruction methods for pulmonary imaging.
- Developed and validated semi-automated DCE-MRI and UTE-MRI pipelines, integrating segmentation, motion correction, parametric modeling and advanced registration and denoising, to quantify global and regional microvascular and structural lung changes in idiopathic pulmonary fibrosis and bronchopulmonary dysplasia.
- Collaborated with clinicians and imaging researchers to interpret imaging-derived biomarkers in the context of pulmonary physiology and disease progression across patient cohorts.
Los Alamos National Laboratory
February 2015 – August 2016
Post-Baccalaureate Researcher
- Developed software to process, analyze and visualize multi-sensor environmental imaging datasets, integrating satellite and spectroscopic data for trace gas and atmospheric modeling.