Generationship Ep. #40, ExperimentOps with Salma Mayorquin of Remyx AI
Jul 24, 2025
Salma Mayorquin, Co-founder of Remyx AI, is dedicated to making tailored machine learning accessible. She dives into the shift from traditional MLOps to ExperimentOps, sharing insights from giants like Netflix, Yelp, and Stripe. Salma discusses the challenges modern AI poses and how companies are integrating foundation models. She also brings to light the value of personal passion projects in AI development, and envisions a future where AI engineering is more accessible and integral to business strategies.
AI Snips
Chapters
Transcript
Episode notes
Foundation Models Revolutionize AI Production
- Foundation models are being integrated across major companies like Yelp, Netflix, and Stripe for various tasks from personalization to payment processing.
- These models enable new product capabilities by leveraging large-scale, flexible AI, reshaping production use cases.
Complexity of Evaluating Foundation Models
- Older ML models were task-specific with clear metrics like accuracy, making evaluation straightforward.
- Foundation models are multipurpose and flexible, complicating evaluation and requiring new approaches to metrics and testing.
ExperimentOps Extends MLOps for AI
- Traditional MLOps is crucial but insufficient alone to handle the complexity of foundation models.
- ExperimentOps is proposed to operationalize knowledge, maximizing learning velocity and product impact in AI experimentation.