LLMs in Production by Christopher Brousseau & Matt Sharp | LLMOps, AI Engineering & Machine Learning

LLMs in Production by Christopher Brousseau & Matt Sharp

LLMs in Production: From Language Models to Successful Products is a practical guide for developers, data scientists, machine learning engineers, and technology professionals who want to move Large Language Models (LLMs) from experimentation into reliable, scalable, real-world applications. Written by Christopher Brousseau and Matt Sharp, this book focuses on the engineering, deployment, evaluation, optimization, and operational challenges involved in putting modern language models into production.

Rather than concentrating only on the theory behind artificial intelligence, the book explores the application layer of foundation models and explains how organizations can design, build, deploy, monitor, and improve LLM-powered systems. It covers the practical decisions that arise when working with large models, including choosing between existing models and building custom ones, preparing suitable datasets, controlling infrastructure and computing costs, evaluating model performance, improving prompts, and maintaining security.

About This Book

LLMs introduce challenges that differ from traditional software and conventional machine learning systems. Their large size, computational requirements, data dependencies, and behavior make careful planning essential. LLMs in Production provides an LLMOps-oriented approach for taking AI applications from initial design through production deployment.

The book explains how to build an effective platform for LLM workloads, work with foundation models, fine-tune existing models, and create applications that take advantage of language-model capabilities while addressing their limitations.

Readers also learn how to approach cost-efficient training, retraining, load testing, deployment architectures, and model optimization for different hardware environments.

What You Will Learn

• Understand the fundamentals of Large Language Models and foundation models
• Decide when to use a pre-trained LLM and when to build or train your own
• Design an effective LLMOps and machine learning platform
• Prepare datasets suitable for LLM training and fine-tuning
• Train foundation models and fine-tune existing language models
• Work with parameter-efficient techniques such as PEFT and LoRA
• Explore reinforcement learning with human feedback
• Develop effective prompt-engineering strategies
• Evaluate LLM performance using practical benchmarks
• Balance model performance, infrastructure requirements, and cost
• Retrain and load-test LLM applications
• Deploy LLM applications to cloud environments
• Understand deployment using Kubernetes
• Optimize models for commodity and edge hardware
• Build AI-powered developer tools and applications
• Deploy smaller language models to resource-constrained devices
• Address security and operational considerations in production AI systems

LLMOps and Production AI

A major focus of the book is the transition from an experimental LLM project to a production-ready AI application. This includes understanding application architecture, data pipelines, compute requirements, model lifecycle management, monitoring, testing, security, and cost management.

The authors explain why LLM applications require a different operational mindset from conventional software. Because large models can be expensive to train and difficult to modify, decisions concerning data, infrastructure, model selection, evaluation, and deployment can have significant long-term consequences.

Training, Fine-Tuning and Model Optimization

The book provides practical guidance on training and adapting language models. Readers can explore the process of working with training datasets, foundation models, and fine-tuning strategies designed to make models more useful for particular applications.

Techniques such as LoRA and PEFT are covered as approaches for making model adaptation more efficient. The book also discusses reinforcement learning with human feedback and other techniques relevant to improving model behavior.

Building LLM Applications

Beyond model development, LLMs in Production focuses on creating applications that use language models effectively. It examines how developers can integrate LLM capabilities into software while designing around their limitations.

Example projects help connect concepts to practical implementation, including building a custom LLM, developing an AI coding extension for VS Code, and deploying a smaller model to a Raspberry Pi.

Deployment and Infrastructure

Production LLM systems can require substantial computing and infrastructure resources. This book explores strategies for scaling machine learning platforms, deploying models to cloud environments, and optimizing models for available hardware.

It also covers deployment considerations involving Kubernetes, commodity hardware, and edge devices, making the material relevant to engineers responsible for taking AI systems beyond local development environments.

Evaluation, Cost and Reliability

Successful AI products require more than an impressive model. Production systems must be evaluated, tested, monitored, and optimized.

The book discusses practical topics including:

• Model evaluation
• Industry benchmarks
• Load testing
• Retraining
• Cost and performance optimization
• Infrastructure planning
• Security
• Production deployment
• Model limitations
• Application reliability

These topics help readers think about LLMs as complete software and machine-learning systems rather than isolated AI models.

Key Topics Covered

• Large Language Models
• Generative AI
• Foundation Models
• LLMOps
• MLOps
• Machine Learning Engineering
• NLP and Natural Language Processing
• LLM Training
• Fine-Tuning
• LoRA
• PEFT
• RLHF
• Prompt Engineering
• Model Evaluation
• AI Application Development
• Kubernetes
• Cloud Deployment
• Edge AI
• Model Optimization
• Load Testing
• AI Infrastructure
• Production Machine Learning
• AI Security
• Cost Optimization

Why Read This Book?

LLMs in Production is especially valuable for readers who already understand Python and basic cloud deployment and want to learn how modern language-model applications are engineered for real-world use. Rather than treating LLMs purely as an academic subject, the book emphasizes practical production concerns and the decisions engineers face when deploying AI systems.

It provides a bridge between understanding what LLMs are and understanding how to successfully use them as part of production software. The combination of LLM fundamentals, application development, infrastructure, fine-tuning, evaluation, deployment, and operational considerations makes it a useful reference for modern AI engineering.

Who Should Read This?

This book is particularly suitable for:

• Data scientists
• Machine learning engineers
• AI engineers
• Software developers
• MLOps engineers
• DevOps and cloud engineers working with AI
• Python developers exploring LLM applications
• Technology professionals building generative AI products
• Students studying modern machine learning engineering
• Developers interested in LLMOps and production AI
• Teams responsible for deploying and maintaining AI applications

The publisher specifically positions the book for data scientists and ML engineers who know Python and the basics of cloud deployment.

Product Details

• Book Title: LLMs in Production: From Language Models to Successful Products
• Authors: Christopher Brousseau & Matt Sharp
• Publisher: Manning
• Language: English
• Genre: Artificial Intelligence / Machine Learning / Software Engineering
• Publication: 2025
• Length: 456 pages
• Print ISBN-13: 9781633437203
• Format: English Paperback / Trade Paperback
• Level: Intermediate to Advanced
• Primary Topics: LLMs, LLMOps, MLOps, AI Engineering, NLP, Machine Learning

Publisher records confirm the 2025 Manning edition, 456-page length, and ISBN 9781633437203.

About the Authors

Christopher Brousseau is a Staff Machine Learning Engineer with a background in linguistics and localization. His work specializes in linguistically informed natural language processing and ML/data product initiatives.

Matt Sharp is an engineer, former data scientist, and experienced technology leader specializing in MLOps and the deployment, management, and scaling of machine learning models in production.

Final Overview

If you are looking for a practical book on building and deploying LLM-powered applications, this is a strong resource for understanding the engineering challenges behind production AI. From model selection and training to fine-tuning, evaluation, infrastructure, deployment, optimization, security, and cost management, LLMs in Production provides a practical roadmap for turning language-model technology into useful and scalable products.