Explore the challenges and solutions for building real-time machine learning pipelines in this 43-minute conference talk. Learn how to handle high-velocity, high-volume data for applications like fraud prediction, predictive maintenance, and customer churn prevention. Discover techniques for online and offline feature engineering, ML calculations in development and production environments, and effective monitoring of AI applications to detect and mitigate drift. Gain insights from real customer case studies and understand how to implement serverless architectures for simplified, high-performance ML pipelines. Delve into topics such as integrated feature stores, Kappa architecture, and MLOps automation to accelerate development and deployment of real-time AI solutions.
Building Real-Time ML Pipelines the Easy Way
Overview
Syllabus
Intro
Most AI Projects Never Make it to Production
Operationalizing Machine Learning is Challenging
Resource Intensive Processes, Data & Org Silos
Serverless Simplicity With Maximum Performance
Accelerate Development & Deployment With an Integrated Feature-Store
Churn Prediction Example: Raw Data Model
Feature Used For The Model (Example)
Implementing A SINGLE Feature Using SQL
Kappa Architecture - Intro
Serverless Stream Processing For Real-Time & Batch
Faster development to production through MLOps & Serverless automation
Rapid Deployment of Real-Time Serverless Pipelines
Glue-less Model Monitoring and Governance
ML Pipeline Example: Predicting Financial Fraud
MLOps for Good Hackathon
Taught by
Open Data Science
Related Courses
-
How to Build a Real-Time ML Pipeline for Fraud Prediction
-
MLOps Automation with Git Based CI-CD for ML
-
Building Real-Time ML Features with a Feature Platform
-
Building Real Time ML Pipelines with a Feature Store
-
Building a Fraud Detection Model with Feature Stores - Includes Shopify Case Study
-
The Challenges of Deploying Real-time AI for Finance and How Open Source Can Help
