Thomas Gerber Interview
summary
Salesforce Einstein builds a platform to democratize AI, ingesting CRM data to predict, recommend and automate while wrestling with data processing, automated model search and meta‑ML. Scaling comes from explicit architectural trade‑offs rather than magic: Spark + Scala pipelines simplify distributed transforms, moving off the JVM to native threads or GraalVM can reduce pain, and hyperparameter search creates many expensive iterative jobs so teams cap compute or accept “good enough.” Practical wins include replacing Hadoop/Cascading with Scalding/Spark and using functional APIs to make pipelines auditable and correct.
AI-written summary · about
related talks
49mImpact on business and society
04mInterview
46mIntroducing Salesforce Genie: A real-time platform for customer magic
47mMetrics Driven Machine Learning Development at Salesforce Einstein
35mHigh-performance data foundation for AI success
03mInterview
14mTony Baer, dbinsight & Doug Henschen, Constellation Research | Databricks Data + AI Summit 2023
22mDoing Better with Data and AI