Nassery-HI1200×1505 82.7 KB

Leemay Nassery

Before deploying an AI model into production, you need to know more than just its accuracy. Will it be fast enough for your users? Will it scale under real-world traffic? Can you trust its decisions in critical scenarios? AI Model Evaluation (Manning Publications) gives you the practical tools and strategies to answer these questions—and more—so you can ship AI systems that actually work in the real world.

What you’ll learn in AI Model Evaluation:

• Build diagnostic offline evaluations to uncover hidden model behaviors
• Use shadow traffic to simulate production conditions safely
• Design A/B tests to measure real business and product impact
• Spot nuanced failures with human-in-the-loop feedback
• Scale evaluations with LLMs as automated judges

Author Leemay Nassery (Spotify, Comcast, Dropbox, Etsy) shares real-world insights on what it really takes to prepare models for production. You’ll go beyond standard accuracy metrics to evaluate latency, user experience, and long-term impact on product goals.

Inside the book:
Each chapter explores a different evaluation method, from offline testing and A/B experiments to shadow deployments and qualitative analysis. Hands-on examples, including a movie recommendation engine, make it easy to apply these techniques to your own AI projects.

• Full details: AI Model Evaluation - Leemay Nassery

Don’t forget you can get 45% off with your Devtalk discount! Just use the coupon code “devtalk.com” at checkout