This episode is for whomever is interested in data workloads for AI particularly event streaming and how those will be used in AI model training and inference.
Andrew Sellers is the Head of Technology Strategy at Confluent. He was previously CTO at QOMPLX and CTO of the US Air Force Academy. In this episode, we dive into technology in the government and what changes for event streaming in the context of AI models. We cover technical aspects of data streaming, customer use cases for AI & event streams, and differences between inference workloads and training workloads for data streams.
Where to Find Andrew Sellers:
LinkedIn: https://www.linkedin.com/in/andrewjsel/
Where to Find Shomik:
Twitter: https://twitter.com/shomikghosh21
LinkedIn: https://www.linkedin.com/in/shomik-ghosh-a5a71319/
Software Snack Bites Newsletter: https://www.shomik.substack.com
Software Snack Bites Podcast: Apple Podcasts, Spotify, Google.
Covers:
(01:52) - Background as CTO of US Air Force Academy
(04:26) - How the Public Sector Optimizes Tech Decisions & Procurement
(7:34) - Enterprise Data Strategies for AI
(9:30) - Why Decoupling is More Important than Latency for Data Streaming Needs
(12:30) - In-Stream Processing
(17:10) - Bi-Directional Streaming
(18:50) - AI Agents Impact on Streaming
(21:10) - Inference Workflow Impact on Streaming
(27:04) - Customer Use Cases
(36:25) - Agent Use for Internal Product Workloads
(38:12) - Real Time Data Attach Rates to Inference
Share this post