Hagay Lupesko on why it is critical for organizations to build their own generative AI models.
Subscribe: Apple • Spotify • Stitcher • Google • AntennaPod • Podcast Addict • Amazon • RSS.
Hagay Lupesko, is VP Engineering at MosaicML, a startup that enables teams to easily train large AI models on their data and in their own secure environment. We discuss the the evolution of cloud based machine learning (from “traditional” ML through LLMs), his experience building machine learning applications at leading technology companies, and the need for companies to build their own custom foundation models.
Interview highlights – key sections from the video version:
- His experience building ML applications at AWS and at Meta
- The different Deep Learning frameworks and their usage in industry
- Why teams will build their own custom foundation models
- The types of foundation models that companies will build
- What type of talent is needed for building custom foundation models
- If you don’t train your own foundation model, you will get disrupted
- For teams who want to build their own foundation models, are there enough open source resources available?
- Scaling Laws and interest in medium sized models
- Advice for people who want to enter the field of applied ML

Related content:
- A video version of this conversation is available on our YouTube channel.
- Building LLM-powered Apps: What You Need to Know
- Jakub Zavrel: Uncovering and Highlighting AI Trends
- Percy Liang: Evaluating Language Models
- Dylan Patel: The Open Source Stack Unleashing a Game-Changing AI Hardware Shift
- Pablo Villalobos: Exhaustion of High-Quality Data Could Slow Down AI Progress in Coming Decades
- Roy Schwartz: Efficient Methods for Natural Language Processing
- Barret Zoph and Liam Fedus: Efficient Scaling of Language Models
- Connor Leahy and Yoav Shoham: Large Language Models
- Mark Chen of OpenAI: How DALL·E works
- Piotr Żelasko: The Unreasonable Effectiveness of Speech Data
- fastdup: Introducing a new free tool for curating image datasets at scale
If you enjoyed this episode, please support our work by encouraging your friends and colleagues to subscribe to our newsletter:
[Image: Model Explosion – by Ben Lorica, using images from Infogram.]