​​About this series

​The series explains the most important topics and pitfalls in the areas of Team, Legal, Finance, GTM and Product Development for AI researchers that build their Frontier AI start-up.

​About this session

​Early-stage AI startups face a classic MLOps dilemma. Build too little and you ship fragile models you can't reproduce or monitor. Build too much and you spend scarce engineering time on infrastructure you don't need yet. In this session, David Sauerwein lays out a minimal viable MLOps stack for teams that use both API-based and self-hosted models: what you need from day one, which signals mean it's time to expand, and how to avoid overengineering along the way. He draws on his experience working in an AI startup, but also working with startups and some of the largest companies in the world while at AWS.

​Meet your speaker

​David Sauerwein is a Principal AI/ML Engineer at AWS Professional Services. He helps the largest AWS customers define AI strategies and deploy AI at scale, and has led the build out of central MLOps and AI platforms for global enterprises. Before that he worked as Data Scientist at the UK Start-Up Faculty. He has a PhD in theoretical physics and worked in quantum computing at the Max Planck Institute for Quantum Optics in Garching. He also regularly writes about AI, digital twins and quantum computing for 36k+ LinkedIn followers.