MLOps.community Context engineering 2.0, Agents + Structured Data, and the Redis Context Engine
74 snips
Dec 16, 2025 Simba Khadder, founder of Featureform and now at Redis, dives into the fascinating world of context engineering for AI. He argues that context, not models, is the real bottleneck for agents. Simba discusses the evolution of feature stores, emphasizing their ongoing value amidst changing ML economics. He introduces a GraphQL-style semantic layer for better data navigation and details how Redis powers these systems with robust capabilities. Plus, he shares insights on how to improve agent functionality by enhancing context access.
AI Snips
Chapters
Transcript
Episode notes
Feature Stores Still Matter At Scale
- MLOps adoption concentrated where ROI is clear: recommender systems and fraud detection drive feature store demand.
- Simba Khadder says feature stores remain valuable and are seeing renewed interest as teams feel real pain points.
Selling Featureform To Redis
- Simba describes selling Featureform to Redis as intensive but a great team and product fit that preserved the team.
- He says Featureform continues and now has more resources to grow inside Redis as a context engine.
Redis As The Online Store Backbone
- Redis has long powered online stores for feature serving and now aims to offer a full-feature-store solution built on Redis.
- Simba frames this as meeting customers where they are and making Redis the best online store option by comparison.