Talks · Workshops · Bio
Speaking: Clinical AI, GenAI Evaluation, and Cascading Architecture
I speak about production machine learning and generative AI in clinical settings: structural label error, custom evaluation of generative AI, and cascading AI architecture. Below are my upcoming and past talks, a short bio, and how to book me.
Short bio for organizers
One line: Maya Malamud is an AI and data science consultant based in Madrid who helps HealthTech teams build clinical AI that works in production and that clinicians can trust.
Longer: Maya Malamud has 15 years of experience in data science and machine learning and holds an MSc in Applied Statistics. She specializes in production ML, generative AI evaluation, and clinical AI. She has two peer-reviewed publications and works in English, Spanish and Hebrew.
Topics
- Structural label error: systematic mislabeling in clinical data and how to detect and correct it.
- Failure-to-Metric: designing custom generative AI evaluation metrics from real production failures.
- Cascading AI architecture: deterministic logic, classic ML, and confidence-gated LLM escalation.
- Personalized modeling from ultra-small clinical data.
Upcoming talks (2026)
The Evolution of the Data Scientist: From Agentic Euphoria to Cascading AI Architecture
hayaData Conference, Israel. About the topic.
Structural Label Error: A Case Study in Women's Cardiac Misdiagnosis
Health Care, Rehabilitation, and Innovation Conference. About the topic and conference page.
Personalized Fetal Growth Monitoring: From Statistical Process Control to a Clinical Model
TLV Lifesciences, Research Track, Tel Aviv. A proof of concept, co-investigated with Dr. Diamanta Benson of The Open University of Israel and Beilinson Hospital's Women's and Maternity Hospital. Conference site.
The Failure-to-Metric Method: Building Evals Your Platform Can't
MLcon and VibeKode Berlin, AI Native Week. About the topic.
Past talks
Don't Let Your RAG Improvise in the ICU
TechGym. Also available as a 40-minute lecture with a full Jupyter notebook. Watch on YouTube.
Building Trustworthy GenAI Systems: Monitoring, Evaluation and Feedback Loops
DSC Europe, November 2025.
Modeling for Ultra-Small Data
hayaData 2025. How personalized modeling delivers clinical-grade insights when traditional big data approaches fail. Watch on YouTube.
Independent Pathways in Data Science: Challenges and Opportunities for Freelancers
Women in Data Science (WiDS) Israel, 7th edition.
Book me
Message me on WhatsApp or by email at MayaM@MalamudAI.com.
Frequently asked questions
What topics does Maya Malamud speak on?
I speak on three recurring themes: structural label error in clinical machine learning, custom evaluation of generative AI (the Failure-to-Metric method), and cascading AI architecture.
What formats do you offer?
I give conference talks, technical lectures, and workshops for healthcare and engineering teams. One of my talks has also been expanded into a 40-minute lecture with a full Jupyter notebook.
How do I book you for a conference or a team workshop?
Message me on WhatsApp, email me, or book a call. Please include the event, the audience, the date, and the length of the slot.
Planning an event or a team workshop?
Send me the event, the audience, the date, and the length of the slot, and I will suggest a topic.
Book a Strategy Call MayaM@MalamudAI.com