MD · MAS · PhD Epidemiology & Translational Science · Computational Precision Health · National Polytechnic Institute · UCSF / UC Berkeley
I'm Eduardo Rodriguez, a physician and PhD epidemiologist who designs clinical AI systems and research pipelines that work because they're built with scientific rigor, not just technical skill.
I work at the intersection of three disciplines I refuse to simplify into one.
I trained as a physician and practiced general and family medicine before pursuing an MAS in Clinical Research and a PhD in Epidemiology & Translational Science with an emphasis in Computational Precision Health.
Today I design and deploy clinical AI systems and research pipelines for groups that need someone fluent in all three languages — clinical context, methodological rigor, and production engineering.
Trained as a physician with hands-on experience in general and emergency medicine. I understand what a clinical finding means before I model it — and I apply that lens to every research question and system I design.
PhD in Epidemiology & Translational Science, MAS in Clinical Research. Every pipeline I build is designed with the same methodological rigor I'd bring to a clinical trial — from study design through endpoint derivation.
I design and deploy LLM systems, automate clinical research workflows, and build data pipelines — end to end, in production, not just in notebooks.
Upturn Health · Deployed · 2025
Designed and deployed a five-phase agentic LLM system for a women's health startup — integrating safety screening, adaptive conversation logic, and clinical protocol alignment.
Stack: Python · LLMs · YAML
UCSF · Research · UCSF
Automated pipelines for oncology research data processing — from EHR extraction through endpoint derivation (OS, PFS, RANO, Genetic variants) to analysis-ready datasets.
Stack: Python · EHR · NGS
Regeneron · Deployed
Built and deployed internal analytical tools used daily by clinical operations teams at Regeneron — a portfolio-level enrollment monitoring dashboard tracking trial status across the company's portfolio, and a subject-level data completeness dashboard. Designed to answer the specific operational question at hand, not to be general-purpose tools.
Stack: Python · Streamlit · SQL
I collaborate with clinical teams, research groups, and health technology companies that need someone who can move between the science, the clinical context, and the implementation. I build, I analyze, and I stay accountable to the evidence.
You have a clinical problem and an AI team that doesn't speak the same language. I bridge that gap — defining requirements from a physician-scientist perspective, designing systems that are clinically valid, and ensuring outputs are interpretable and defensible.
Clinical research teams spending weeks on data wrangling, endpoint derivation, or manual extraction. I design automated pipelines — LLM-powered and traditional — that accelerate timelines without sacrificing rigor.
Genomic profiling of 205 IDH-mutant astrocytomas revealed that RAS-MAPK pathway alterations are enriched in recurrent, high-grade gliomas and associated with inferior survival, implicating this pathway as a key driver of malignant transformation and a potential therapeutic target.
GPT-4 and Claude-2 were both capable of generating clinically valid molecular subtyping decision trees across five cancer types, though adding expert-curated guideline text offered only modest accuracy gains — suggesting LLMs hold real promise for distilling oncologic diagnostic frameworks but require careful oversight before clinical deployment.
