What is a Forward Deployed Engineer?

A forward deployed engineer (FDE) is a software engineer who works directly with a customer’s teams, data and systems to turn a business problem into a working solution in production, and stays involved until people use it. Instead of building one capability for many customers from a distance, a forward deployed engineer works close to one customer’s people, learns how the work is actually done and shapes the software around it.

The role has grown quickly with AI. US Google searches for “forward deployed engineer” rose from about 8,100 a month in October 2025 to 40,500 in September 2026, and searches for “what is a forward deployed engineer” grew more than sixfold over the same period. Most writing about the role treats it as a job title. This guide covers both sides: what the engineer does, and what a company gets from working with one.

Sitation offers forward deployed engineering for product data and AI. The short definition is also in our product data glossary.

Jump to: What they do · Skills · Compared with other roles · Why AI projects need them · When you need one · Product data and commerce · Working with one · FAQ

What does a forward deployed engineer do?

A forward deployed engineer turns a problem described in business terms into software that people use every day. The work moves through four stages, repeated as the solution grows.

1. Learn how the work is done. Sit with the people who do it, read the real data and trace the systems involved. The aim is to find the details that never make it into a requirements document.
2. Decide what to build. Turn what they learned into options, and agree the first piece of work with the business and technical owners. Keep the scope small enough to test with real users.
3. Build and release. Write production software: integrations, data pipelines, internal tools, AI workflows and the permissions and checks around them. Release in increments and adjust as people use it.
4. Hand over or stay. Document what was built and transfer it to the team that will own it, or keep improving it as an ongoing service.

A typical week

A forward deployed engineer’s week mixes conversations and code. It might include a working session with a merchandising team to watch how they fix product records, an afternoon building an integration, a review of an AI agent’s answers with the support lead and a short demo of what changed. The engineer also carries what they learn back to their own company, so problems that keep coming up become better tools.

Skills a forward deployed engineer needs

The role asks for a wider mix than most engineering jobs. Strong forward deployed engineers combine:

• Broad engineering. APIs, data modeling, integration, cloud services and, increasingly, AI tools such as language models, retrieval and agents. They build production software, not only prototypes.
• Data fluency. Reading messy data, finding where sources conflict and designing the rules that keep it clean.
• Domain understanding. Enough knowledge of the customer’s business to see why a process works the way it does, for example how a retailer sets up new items or how a distributor handles substitutions.
• Communication. Explaining trade-offs to people who are not engineers, and turning their answers into decisions.
• Judgment about scope. Choosing the smallest useful thing to build first, and saying no to work that will not change the result.
• Ownership. Treating the result in use as the measure of success, not the number of tasks closed.

In product data work, domain understanding means knowing product information: attributes, taxonomies, variants, digital assets and the channel and retailer requirements they feed.

Forward deployed engineer vs other roles

The title overlaps with several familiar roles. The differences are in who decides what to build, who builds it and who is responsible once it is in use.

Why AI projects need forward deployed engineers

AI widens the gap between a demo and daily use. A model can produce a convincing answer in a meeting. Getting the same quality every day depends on details that live inside the business:

• The data. Which product attributes are approved, which sources conflict and which records an agent may read or change.
• The rules. Brand voice, regulated claims, retailer requirements and the exceptions every catalog has.
• The review points. Who checks the output, when a person must approve a change and how errors are caught.
• The systems. Connections to the PIM, ERP and commerce platforms where the work actually happens.
• The people. Whether the team that owns the workflow trusts the result enough to use it.

Few of these details make it into a requirements document. They come out when an engineer sits with the merchandiser who fixes product records, or the support agent who answers questions about fit and compatibility. A forward deployed engineer finds them early, builds them into the solution and adjusts as the team uses it.

When you need a forward deployed engineer, and when you don’t

The forward deployed model earns its place when the problem is real but the solution depends on details nobody has written down yet. That describes most AI work today, and many integration and data quality problems. Common signs:

• An AI pilot works in a demo but not on your real data.
• Several teams own parts of the process, and nobody owns the whole result.
• Requirements keep changing as people see working software.
• Your engineers know the systems but not the business process, or the other way around.

You probably do not need one when requirements are clear and stable, when you need a recommendation before you commit, or when your team knows exactly what to build and needs more hands. A defined project, a consultant or added capacity will be simpler.

Forward deployed engineers in product data and commerce

In product information work, the role usually takes two shapes. A forward deployed AI engineer connects agents and generative AI to approved PIM data, sets permissions and review points, and measures whether the output is good enough to use. A forward deployed PIM architect works alongside merchandising, eCommerce and IT teams on the data model, governance and platform decisions that later AI work depends on.

Both need product data expertise as well as engineering. Many AI problems in commerce are data problems underneath. An agent that recommends the wrong product often has a missing attribute or a conflicting source behind it, not a weak model.

Typical work

• AI content and enrichment grounded in approved product data, with human review. See our guide to AI for product content and enrichment.
• AI agents that answer support and sales questions from PIM and MDM data, as in Sitation’s Managed Agents.
• Integrations between PIM, MDM, DAM, ERP and commerce systems. See our integration expertise.
• Readiness for AI shopping assistants, where product data must be accurate enough for an agent to compare and recommend. See what agentic commerce is.

On Salsify, Akeneo and Stibo Systems

The model is the same on any platform, and the details differ. On Salsify the work often centers on channel syndication, enhanced content and the Salsify API. On Akeneo it often involves families and variants, Akeneo’s APIs and apps that extend the PIM. On Stibo Systems STEP it spans product, customer and supplier data, with governance rules and integrations across the business. See how Sitation works as Salsify forward deployed engineers, Akeneo forward deployed engineers and Stibo Systems forward deployed engineers.

How to work with a forward deployed engineer

The model works best when the customer sets it up to succeed. Six practices make the biggest difference:

1. Define the result, not the tasks. Describe what should be true when the work is done, such as fewer manual corrections or an agent that answers fit questions correctly.
2. Give access early. Access to the people, data and systems in the first week saves weeks later.
3. Name a business owner. One person who can make decisions about the process, not only approve the budget.
4. Review working software often. Short, regular reviews keep decisions moving and catch misunderstandings while they are cheap to fix.
5. Plan the handover from the start. Decide who will own the result, and make documentation part of the work.
6. Measure use, not output. Count the decisions made faster, the corrections avoided or the questions answered correctly, not the features delivered.

Two signs show it is working. Your team makes decisions faster, because the engineer brings options grounded in how the work is done. And people use the first version early, instead of waiting for one large launch.

How Sitation’s forward deployed engineers work

Sitation’s forward deployed engineers are AI engineers and PIM architects who work inside our clients’ teams on product data, integration and AI, from part of one engineer’s week to a full Sitation team. They follow four steps: frame the work, build in increments, validate and release, then transfer or continue.

When the result should keep running, the work can continue as PIM managed services. For a new platform, start with PIM, MDM and DAM implementation. For AI use cases, see our AI digital transformation services.

Frequently asked questions