A day in the life of a machine learning data scientist at JPMorgan Chase
summary
JPMorgan builds OmniAI on Amazon SageMaker as a hybrid deployment that keeps control plane and identities on‑prem while running training on AWS; the platform adds JPMorgan KMS backed by CloudHSM, log forwarding, and automated VPC/subnet/role setup. Demo trains a logistic regression (Scikit‑learn) on an FDIC dataset (~178,000 rows) on an m5.large, produces a tarred model, and the program reports success. An internal claim is $150M in cost savings to date, and the OmniAI SDK hides buckets/keys so data scientists focus on models.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 2:02Requirements
- 3:33Agenda
- 4:46JPMorgans scale
- 6:11Use cases
- 8:37Barriers
- 12:57OmniAI
- 13:17OmniAI Platform
- 17:32Demo
- 22:07Data set
- 22:56SageMaker Omni SDK
- 26:37Other Omni SDK features
- 27:43Results
- 33:37Next steps
related talks
04mCustomer Spotlight
43mMachine Learning Ops: DevOps for Data Science
55mMachine Learning and AI in Java on AWS
19mDrive transformation through machine learning
31mLet Devs Be Devs: Abstracting Compliance + Reliability to Accelerate JPMCs Cloud Deployments
19mUsing Amazon SageMaker to develop and deploy ML models
37mStreamlining Machine Learning Operations
29mMachine learning in production