Astrid
Bernaga

Physician & Applied AI Specialist

Building at the intersection of clinical medicine, machine learning, and digital health. CTO at Unephra. MD + MSc Applied AI.

Astrid Bernaga
Physician Applied AI CTO · Unephra NLP Research ACL 2025 Digital Health Data Science Healthtech Physician Applied AI CTO · Unephra NLP Research ACL 2025 Digital Health Data Science Healthtech

I'm a physician and applied AI product leader working at the intersection of clinical medicine, machine learning, and digital health.

Currently serving as CTO at Unephra, where I lead data infrastructure, clinical pilot design, and the application of large language models to translate biomarker data into meaningful health insights.

My background spans clinical practice, AI/ML development, and startup product operations, with a focus on building systems that make healthcare more proactive, personalized, and accessible.

MD from Universidad de Guadalajara · MSc Applied AI from Tecnológico de Monterrey · Co-author, ACL 2025 in collaboration with UTSA.

What I do

01
AI & Health Tech
Building ML and LLM systems that translate clinical data into actionable health insights.
02
Clinical Research
Designing pilots and studies that bridge clinical workflows with digital health systems.
03
Product & Startup Ops
Leading cross-functional execution from concept to deployment in early-stage environments.
04
Speaking & Research
Presenting at international AI and health conferences. Published in ACL 2025.

Selected projects

Current · CTO
Unephra Health Monitoring Platform
Leading data infrastructure and LLM integration for a passive urine-monitoring device enabling at-home health monitoring.
Research · ACL 2025
Dialect Bias in LLMs: Multi-Agent Framework
Co-authored peer-reviewed research on reducing dialect bias in privacy policy QA systems using multi-agent LLMs. Accepted to ACL 2025 Main Conference.
ML · Fundación Carlos Slim · Apr-Jun 2024
Cardiovascular Risk Prediction Model
Built a clinical risk prediction prototype in Python (Pandas, scikit-learn, TensorFlow) grounded in cardiovascular guidelines. Translated clinical requirements into model features, evaluation criteria, and validation workflows, then applied medical expertise to analyze outputs and drive iterative improvements.

Live tools

Talks & conferences

Let's connect

Open to collaborations, speaking invitations, advisory roles, and conversations about AI in healthcare.

hi@astrid.mx
Book a session →