How Rocket Companies modernized their data strategy on AWS
summary
Rocket Companies moves from a monolithic Hadoop lake to an AWS stack using Aurora, Dynamo, Redshift, Lambda, EMR, Glue Catalog, Athena, SageMaker Studio, Bedrock and Kendra and ingests petabytes from 150 sources. Ingestion and ETL that once took months now run in under two hours, transient EMR clusters cut compute from 36 nodes to two, saving ~40% and $3M, managed spot training lowers training cost up to 90%, and the platform powers 3.7 billion automated AI decisions per year while freeing data scientists to spend 80% of time on innovation using SageMaker Studio and shared metadata in Glue Catalog.
AI-written summary · about
chapters (18)
- 0:00Intro
- 0:44Agenda
- 1:48Mortgage in a Box
- 2:31Rocket Mortgage
- 3:41Biggest challenge
- 4:34Our strategy
- 5:17Our secret sauce
- 7:19AI ready data
- 9:41Data Lake and Big Data
- 10:10Data Science Project
- 13:08modernized architecture
- 13:49ingestion
- 14:08EMR
- 14:34Transient EMR
- 15:01Data Science Platform
- 16:02Stagemaker
- 18:46Whats next
- 20:07Use cases
related talks
28mTurn Data Into Insights
45mAccelerated analytics: Building the next-gen data platform for Hertz
28mInnovations in AWS analytics: Data processing
29mSimplifying data flows
22mTurn Data Into Insights
33mAnalytics at scale: Migrating Intuit to AWS with cloud-native tools
44mBest practices for analytics and generative AI on AWS(ANT302)
22mFireside Chat with AWS