As more businesses shift from batch to stream processing to meet the needs of today’s world, so does machine learning. Real-time ML is rapidly expanding due to an exponential increase in both data volume and velocity. This shift, however, is not without its own set of challenges. Building real-time ML pipelines can be difficult and exhausting, calling for skills in both ML and streaming processing. Given the high initial investment in infrastructure, many companies are reluctant to consider the overall benefits. Moreover, most of the traditional ML methods are designed to process data in batches, which requires a profound paradigm shift. 

In this panel, participants will share their thoughts on the increasing importance of real-time ML and its overall benefits. They’ll discuss typical use cases and the challenges of building real-time ML pipelines. Furthermore, they’ll look into different tools and frameworks that can help apply ML to streaming data and explore different ML architectures, illustrating how those can solve the real-time problems of today. Finally, we’ll invite our participants to share their perspectives on unifying batch and stream processing by using a streaming-first infrastructure, streaming feature stores, and other emerging trends. 

Panelists

Chip Huyen, Co-Founder at Claypot AI
Abhay Bothra, Co-Founder & CTO at Fennel
Rohit Agrawal, Engineering Manager at Tecton
Yingjun Wu, Founder and CEO at RisingWave Labs