Suresha-HI1200×1505 74.9 KB

Suhas Suresha, Dewang Sultania

Most AI projects don’t fail because of the model. They fail in everything around it.

We’ve just released Designing AI Systems, a book that focuses on that missing layer—the parts you end up building again and again once you move past a demo.

Instead of stopping at “call an API and get a response,” this book walks through how to build the system that makes those calls useful in practice. You’ll put together a shared AI platform step by step, the kind teams rely on when they have more than one experiment running and need consistency.

That includes:

• connecting to multiple model providers without rewriting your app each time

• keeping track of conversations and context over time

• bringing in your own data so outputs aren’t generic

• letting AI systems use tools while staying within clear boundaries

There’s also a strong focus on what happens after the first version ships. How do you keep costs under control when usage grows? How do you route requests, measure quality, and catch failures before users do? Those questions are part of the build, not an afterthought.

Each chapter includes working Python code, so you’re not left guessing how the pieces fit together. By the end, you’ll have a reference system you can adapt, extend, or pick apart for your own projects.

If you’ve been building with LLMs and keep running into the same infrastructure questions, this book should feel familiar in a good way.

• Full details: Designing AI Systems - Suhas Suresha, Dewang Sultania

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