What Is a Forward-Deployed Engineer, and When Does Your Company Need One?
TL;DR
- A forward-deployed engineer (FDE) embeds with your team to build and ship a production system in your environment, then helps your team take it over.
- FDEs differ from consultants and solutions architects because they are accountable for implementation, not just recommendations or technical designs. Complex AI projects often need all three roles.
- In AI projects, FDE core work is around the model: connecting trusted data, integrating business systems, meeting security requirements, setting up evaluation and observability, and managing cost.
- Most AI-vendor FDEs are deployed at larger enterprises. SMB and mid-market companies are often better served by an AWS partner FDE team than by hiring a specialized in-house role or relying on a model provider’s deployment team. The right partner brings broader data, application, security, and infrastructure expertise for the project’s duration.
A successful FDE engagement leaves the customer self-sufficient. Documentation, runbooks, evaluation processes, cost controls, and knowledge transfer are part of the delivery plan from the start.
What is a forward-deployed engineer?
Definition: A forward-deployed engineer (FDE) is an engineer at a technology vendor who works embedded with a customer's team, writing production code on the customer's own infrastructure and data until a complex technology delivers results for the business. The FDE is accountable for a running system the customer can operate after the engagement ends, which is what separates the role from consultants, who deliver recommendations, and solutions architects, who typically deliver designs.
Brief history: The FDE role predates generative AI. Palantir created it in the early 2010s, when the company found that its data integration and analytics software couldn't deliver value to intelligence agencies, and later to customers such as Airbus and the CDC, until someone connected it to each organization's fragmented data and workflows. Rather than leave that work to customers or outside consultants, Palantir sent its own engineers, internally called "Deltas," to work on-site, often three or four days a week, and fed what they learned back into its products. The logic of the role has held since: when a vendor sells technology too complex to deliver value out of the box, the vendor places engineers at the customer to close the gap.
FDEs for AI: Generative AI pilots are easy to launch, but few make it into production. Foundation models are general-purpose, and their business value depends on each company's data, permissions, and workflows, so AI vendors have adopted the FDE approach. When OpenAI started an FDE team in 2025, job postings for the role rose more than 800% in the months following, and in June 2026 AWS committed $1 billion to a forward-deployed engineering unit of its own. For engineering and technical leaders, it’s important to understand how an FDE differs from the consultants and architects you may already work with, what the work looks like on an AI use case, and how to get an FDE for your own AI project.
How a forward-deployed engineer differs from a consultant or a solutions architect
Forward-deployed engineering, consulting, and solutions architecture are distinct roles, but complex AI projects often require all three. A team needs someone to clarify the business problem, someone to design the technical approach, and someone accountable for building and integrating the production system.
Stratus10 has combined those disciplines in client engagements long before “forward-deployed engineer” became a widely used title. For AI projects, that means we can help identify the right use case, design the solution, build it in your AWS environment, and prepare your team to operate it.
With all the buzz around FDEs, let’s break down how this relatively new role at complex technology companies compares to more-understood titles. Brief definitions before comparing them:
- Forward-deployed engineer: A software engineer embedded with one customer’s team who builds and ships the production system in the customer’s environment, then hands it over.
- Consultant: An advisor, at a strategy firm, a large professional services firm, or working independently, who assesses the current state and recommends a direction. Implementation is often handed to the customer’s team or another vendor.
- Solutions architect: A technical designer who translates business requirements into a cloud architecture. AWS solutions architects work as part of a customer’s AWS account team and advise many customers at once. Solutions architects at AWS partners design the solution for a specific engagement and scope the work before delivery begins.
The differences lie in what each role is accountable for and when their engagement ends.
How a Forward-Deployed Engineer Compares to Other Roles
Recent Posts
Forward-Deployed Engineer
Consultant
Solutions Architect
Delivery and handoff:
Delivers a working production system and hands it over when your team can run and extend it.
Delivers recommendations and a roadmap, usually concluding when the agreed deliverable is accepted.
Delivers a reference architecture and design guidance, usually concluding when the design is approved.
How they engage:
Works inside your team, AWS environment, codebase, and data to build and deploy the solution.
Works with leadership and internal stakeholders to assess needs, evaluate options, and guide decisions.
Translates business requirements into a technical design and advises on the architecture. May build prototypes or samples.
What success looks like:
A production system that is adopted by the business and can be operated by the client team.
Useful, well-supported guidance that helps the organization make a sound decision.
A design that is technically sound, practical to implement, and aligned with the business need.
The roles are easiest to understand separately, but the strongest delivery model combines them. The consultant frames the opportunity, the solutions architect designs the approach, and the forward-deployed engineer carries it through implementation and handoff.
What a forward-deployed engineer does during an AI project
For AI engagements, Stratus10 combines forward-deployed engineering with consulting and solutions architecture across five phases: discover, prioritize, prototype, build and integrate, and evaluate and operate.
The first weeks: finding the workflow worth changing
The first job is to sit with the people who do the work, not only the executives sponsoring the project, and find a workflow where AI measurably changes cost, speed, or quality. Candidate use cases get compared on expected value, technical feasibility, risk, cost, and time to production. A use case that looks impressive but touches data nobody can legally use, or depends on a system with no API, drops down the list quickly.
Data readiness gets assessed at the same time: where the relevant data lives, who owns it, how much is structured versus unstructured, and what can't leave which boundary. Our AI readiness assessment walks through those checks with a 20-question scorecard.
Model selection can also begin in the first weeks too. Our view is that production work should land on models that are proven and available inside your own AWS account through Amazon Bedrock. Before anything gets built, the engineer and the customer agree on success metrics and assemble an evaluation set from the customer's real tasks and documents.
Building on your systems, with your data
Building inside the customer's AWS account, under the customer's IAM policies, networking, logging, and guardrails, means security review happens once and the system belongs to the customer from the first day. Pulling a prototype out of a vendor's sandbox and rebuilding it to meet internal security requirements is one of the most common reasons pilots never reach production.
Most useful business AI has to reach both structured and unstructured data. For Peachjar, a K-12 platform for digital school flyers, Stratus10 built two Amazon Bedrock knowledge bases: one on Amazon OpenSearch Serverless for flyer PDFs and images, with automated Lambda ingestion, and one on Amazon Redshift that turns natural-language questions into SQL over engagement metrics. Non-technical administrators can now ask questions that span both content and performance data. Our guide to Amazon Bedrock data ingestion best practices covers the pipeline decisions behind builds like Peachjar's.
Forward-deployed engineers also spend a large share of their time on the systems around the AI. An agent can only act on applications it can reach, and older applications often need APIs, containerization, or data cleanup before an AI workload can use them. AI-assisted development tools speed up that integration and application modernization work considerably, and an experienced FDE uses them as part of the job.
A prototype may come together quickly, but hardening it takes longer. The FDE evaluates it against the agreed test set, observability, cost tagging and budgets, fallback paths for when the model is wrong, and human review at the points where errors are expensive.
Handing it over so your team can run it
Handoff starts at the beginning of the project. Pair programming, documentation, runbooks, and an evaluation suite the customer's engineers can rerun on their own all build toward the point where the customer no longer needs the forward-deployed engineer for that system.
The handoff includes the documentation, runbooks, evaluation processes, cost controls, and governance your team needs to operate and improve the solution over time. Per-task cost tracking, budget alerts, and defined human-review points provide ongoing visibility and oversight.
When to hire an AI engineer, and when to bring in a partner FDE
Companies have four realistic ways to get forward-deployed engineering for an AI project.
Hire forward-deployed engineers in-house. Hiring fits when AI is core to your product and you have a multi-year roadmap that justifies permanent headcount. The role is scarce and expensive to recruit for, and a single hire rarely covers data engineering, application integration, security, and AWS infrastructure at once.
Work with AWS's FDE unit. AWS's $1 billion unit, launched in June 2026, places pods of five to six engineers inside customer organizations to build agentic AI systems, with the goal of leaving customers self-sufficient within weeks. Early customers include the Allen Institute, Cox Automotive, the NBA, the NFL, and Ricoh. The unit fits large enterprises with a strategic AWS relationship, and availability for mid-market companies hasn't been established yet.
Work with a model provider's FDE team. OpenAI and Anthropic both run deployment teams that work directly with customers. Like Palantir's original FDEs, vendor engineers are there to make their company's product succeed in your environment, and the work naturally centers on that provider's platform. That's a good fit when you've already standardized on one provider.
Work with an AWS partner that has forward-deployed engineers. A partner fits most SMB and mid-market companies. You get a team with data, application, security, and infrastructure depth for the length of the project, and they build inside your own AWS account. AWS is now investing in expanding this same model with the launch of its Partner-Led FDE program in June 2026.
What to look for in a forward-deployed engineering partner
When you're evaluating a partner for forward-deployed AI work, these questions separate engineering firms from advisory firms:
- Do they ship code in your environment?
Ask to see production systems they've built in customer AWS accounts, not slide decks or reference architectures.
- Do they have AI production references on AWS?
Look for case studies involving Bedrock, data pipelines, and integration with real business systems.
- Is their depth broader than the model?
Most of the work in a production AI system is data engineering, security, cost management, and integration with existing applications.
- How are they validated by AWS?
Partner tier and AI-related competencies tell you AWS has reviewed their work.
- Is the handoff in the statement of work?
A defined handoff plan and success criteria, written into the contract, signal a partner that expects to leave you self-sufficient.
- Will they recommend a proven model over the newest one?
A partner willing to say "not yet" about a new release is protecting your timeline and budget.
- Can they help with funding?
AWS offers proof-of-concept funding for generative AI use cases, and an experienced partner can tell you whether your project fits and handles the application.
Stratus10 is an AWS Advanced Tier Services Partner, and our forward-deployed engineers build AI systems inside our customers' AWS environments, alongside their teams, through to production and handoff.
Get support to plan and build your AI use case
Connect with Stratus10 about an AI use case that needs more than a recommendation or a prototype. Our team brings strategy, architecture, and hands-on engineering together to build and hand over a production-ready solution in your AWS environment.
If you have an AI use case in mind, or a pilot that hasn't made it to production, talk to an AI specialist at Stratus10. We'll look at the workflow, the data behind it, and what it would take to get it running in your AWS environment. We can also tell you whether the project qualifies for AWS proof-of-concept funding, which can cover up to $25K of a generative AI POC depending on the project and current AWS program terms.
AI Readiness Assessment
Not sure if your specific AI use case is ready for a proof of concept? Score your readiness, identify any bottlenecks, and see what groundwork to address before or alongside implementation.
Take the AI Readiness Assessment →
Frequently asked questions about forward-deployed engineers
Complex platforms often can't deliver value until they're connected to a customer's specific data, systems, and workflows. Embedding engineers gets the product working in production faster, and what those engineers learn in the field feeds back into the vendor's product.
A forward-deployed engineer works with the customer's business and technical teams to understand a workflow, then writes the code, data pipelines, and integrations to improve it, directly in the customer's environment. As the project matures, more of the work shifts to evaluation, cost and security hardening, and training the customer's engineers to take over.
A solutions architect typically designs a system and advises on it, often across many customers at once. A forward-deployed engineer builds the system inside one customer's environment and stays until that customer can run it.
Engagement length depends on the scope of the use case and the state of the customer's data and systems. AWS describes its own FDE pods as aiming to leave customers self-sufficient within weeks, and a well-scoped first use case with accessible data can follow a similar timeline. Broader programs often run as a series of shorter engagements, one use case at a time.
Yes. AWS launched a $1 billion Forward Deployed Engineering unit in June 2026, starting with large enterprise customers, along with a Partner-Led FDE program that extends the model through AWS partners.