The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Real-Time ML Workflows at Capital One with Disha Singla - #606
16 snips
Dec 19, 2022 
Disha Singla, Senior Director of Machine Learning Engineering at Capital One, leads efforts to democratize ML across the company. She discusses the creation of reusable workflows and libraries designed for citizen data scientists, enabling quicker decision-making. The conversation dives into the implementation of real-time ML, balancing the need for speed in insights with regulatory compliance. Disha also shares strategies for securing executive buy-in by demonstrating ROI and emphasizes the importance of rigorous testing and observability in ML models.
AI Snips
Chapters
Transcript
Episode notes
Forecasting for Hybrid Work
- Capital One's Workplace Solutions team uses Disha Singla's platform for forecasting.
- They predict how many people will return to the office to optimize kitchen stocking.
Transaction Fraud Detection
- Capital One's third-party fraud team uses Singla's platform to detect transaction anomalies.
- The platform automatically generates rules to mitigate losses and reduce customer friction.
Model Development Rigor
- Treat production models like software, applying rigorous software development lifecycle practices.
- Focus on data quality testing, including checks for data types and missing values.