From on-site problem discovery to AI-native production delivery, systematically master the methods, tools, and practices of forward deployed engineering.

About This Book

Forward Deployed Engineering, abbreviated as FDE, is translated in this book as "forward deployed engineering". It is not traditional pre-sales, outsourced implementation, or one-off consulting delivery, but an engineering practice that embeds engineers in real business environments, using software, data, AI, platforms, and operational methods to solve complex organizational problems. The goal of an FDE is not to produce a solution document, but to turn ambiguous problems into production systems that are runnable, observable, governable, and sustainably evolvable.

FDE has expanded from the organizational practice of a few companies into an important delivery model for enterprise AI and digital transformation:

  • Palantir's FDE practice brought this role into industry view and provides AI FDE in the Foundry documentation.
  • OpenAI launched the OpenAI Deployment Company, emphasizing engineers embedding in organizations, identifying high-impact scenarios, and re-architecting key workflows.
  • Accenture and Microsoft established an FDE practice focused on scaling enterprise AI.

This book is aimed at engineers, technical leads, product managers, Solutions Architects, platform engineers, AI application developers, and enterprise digital transformation leaders who want to systematically master FDE.

Target Readers

This book is suitable for the following readers:

  • Engineers with a software engineering foundation who want to enter the fields of FDE, solutions engineering, AI delivery, platform engineering, or enterprise technical consulting.
  • Technical leads who are deploying AI, data platforms, cloud-native platforms, or business automation systems into enterprise environments.
  • Managers who need to understand the boundaries between FDE and pre-sales, consulting, SRE, platform engineering, and customer success.
  • Teams that want to build replicable delivery capabilities with open-source tools, including users of the Kubernetes, GitOps, IaC, OpenTelemetry, Backstage, Temporal, Dagster, LangGraph, LlamaIndex, and MLflow ecosystems.
  • Teams responsible for high-security, high-compliance, high-complexity on-site delivery, such as in finance, manufacturing, healthcare, energy, the public sector, and supply chain scenarios.

What You Will Learn

After reading the entire book, readers should be able to develop four types of capabilities.

First, understand the essence of the FDE role. Readers will be able to distinguish the responsibility boundaries between FDE, Sales Engineers, Solutions Architects, professional services, platform engineers, SRE, and AI engineers, and understand why the core of FDE is on-site discovery, engineering delivery, production value, and product feedback.

Second, master the delivery method from problem to system. The book will explain how to enter the customer site, identify real bottlenecks, establish domain language, design success metrics, split prototypes, pilots, and production systems, and manage scope, risk, change, and knowledge transfer.

Third, build a production-grade engineering foundation. Readers will learn core capabilities such as containers and Kubernetes, GitOps, infrastructure as code, CI/CD, software supply chain security, data products, semantic layers, zero trust, observability, SLOs, and incident response, and understand their practical role in FDE projects.

Fourth, master AI-native FDE practices. The book will cover LLMs, RAG, agents, evals, LLMOps, prompt engineering, tool calling, model deployment, security safeguards, and human-machine collaboration boundaries, with a focus on how to connect AI capabilities to customer data, systems, permissions, and real workflows.

Reading Paths

The table below is a quick path, not a license to skip dependencies. Before a real project goes to production, you should check back on data, security, LLMOps, operations, and the delivery evidence package template to confirm that key evidence, owners, rollback, and takeover conditions are complete.

FDE / Solutions Engineer

1 -> 2 -> 3 -> 4 -> 7 -> 8 -> 10 -> 11 -> 12 -> Appendix templates

Role boundaries, on-site discovery, architecture trade-offs, data and security, production launch, operations closed loop

Platform / DevOps Engineer

5 -> 6 -> 8 -> 12 -> 13

Cloud-native, GitOps, supply chain security, observability, platformization

AI Application Engineer

7 -> 8 -> 9 -> 10 -> 11 -> 12 -> 16 -> Appendix templates

Data ingestion, permissions and security, RAG, agents, evals, LLMOps, AI FDE

Technical Lead / Product Owner

1 -> 4 -> 11 -> 12 -> 14 -> 15 -> 16 -> Appendix templates

Delivery models, takeover criteria, industry cases, organizational capability, long-term value

Related Books

Book Structure

The book is divided into four parts with sixteen chapters, plus appendices.

Part One, "Getting Started: Understanding FDE," establishes the basic concepts. It explains the origins of FDE, role boundaries, on-site problem discovery, architecture trade-offs, and delivery collaboration.

Part Two, "Engineering: Cloud-Native and Automated Delivery," covers the foundational capabilities needed for productionization. It covers cloud-native runtime environments, automated delivery, data integration, security and compliance, and governance.

Part Three, "Production: AI-Native Systems," focuses on enterprise AI implementation. It covers LLM applications, RAG, agents, evals, LLMOps, from prototype to production systems, and on-site operations.

Part Four, "Practice: Scaling and Industry Implementation," discusses how to turn a single delivery into reusable capability. It includes the tool ecosystem, industry cases, organizational capability, career development, and future trends.

The appendices include a glossary, references, a tool index, the delivery evidence package template, comprehensive hands-on practice, the Chapter 14 attestation ledger, and the fast-changing facts verification table, making it easy for readers to quickly consult, verify, and reuse during reading, review, and project practice.

Reading Online

The online version can be read on GitBook: A Practical Guide to Forward Deployed Engineering.

Downloading Offline Versions

The official PDF version can be downloaded from the GitHub Releases page.

Preview versions auto-updated from the default branch can be downloaded directly as PDF, single-file HTML, and SHA256SUMS. These files are overwritten as the mainline updates and do not represent official releases; official tagged versions additionally include build provenance.

Local Reading and Building

Install mdPress:

Local preview:

Check Markdown links and code block fences:

Build the static site, PDF, and ePub:

Generate a strict single-file HTML (requires Pandoc 3.5, Chrome, and Node.js):

CI and Releases

This repository is configured with GitHub Actions:

  • CI: Runs tests and project checks on pushes to the main branch, Pull Requests, and manual triggers; builds and verifies the PDF, single-file HTML, and checksums.
  • Update Preview Publications: First generates and verifies the PDF, HTML, and checksums in a read-only build job, then a separate write job updates the preview-pdf preview Release.
  • Auto Release: On pushes of v* tags, builds the PDF, HTML, and checksums; a separate release job generates build provenance attestations and uploads the GitHub Release; manual triggers only generate verified artifacts and do not write an official Release.
  • Dependabot auto-merge: Auto-merge is enabled only for low-risk patch and minor version updates of GitHub Actions, and requires that required checks are configured on the target branch.

Last updated