Forward Deployed Engineers: What's Behind the New AI Job Title

Billions in investments in a new professional role: Microsoft, Amazon, OpenAI, and Anthropic have all announced their own units for Forward Deployed Engineering within just a few weeks. Microsoft alone is investing approximately 2.5 billion U.S. dollars and plans to deploy 6,000 engineers and industry experts to work with customers; OpenAI, together with investors, has founded the Deployment Company; and Anthropic’s announced joint venture is explicitly aimed at small and medium-sized businesses.

The goal is to further expand the productive use of AI and bring so-called Frontier AI to customers. “Frontier AI” refers to the most powerful AI models currently available, which solve complex problems and are at the cutting edge of technological development.

Behind all these initiatives lies the same model: Forward Deployed Engineering. We explain what the term means, why major providers are now embracing Forward Deployed Engineering, and why this is of interest to companies, as well as employers and job seekers in the tech industry.

First off: FDEs haven’t just emerged recently, nor did they come about only since AI became the dominant topic. However, the current wave of attention was triggered by challenges in implementing AI projects within companies. It has brought visibility to the role of the Forward Deployed Engineer and, in our view, could spark a shift in thinking when it comes to planning complex software.

A Forward Deployed Engineer (FDE) is a software developer who works directly at the client’s site and within the client’s processes to implement complex software projects. For example, they analyze the infrastructure and possibilities for AI use cases directly at the client’s site, build the solution using the company’s data and systems, and remain responsible for it until it delivers measurable results.

The term was popularized in the early 2010s by Palantir, which has since been sending its developers to its clients’ organizations to integrate its products and adapt them to their processes. At Palantir, the focus was long on this integration work: adapting existing software products to the clients’ established systems, building interfaces, and migrating data.

A Forward Deployed Engineer (FDE) who implements AI applications is tasked with integrating models, agents, or RAG systems into a company’s operations and bringing them to production readiness. This new role has emerged from the current demand for AI solutions and explains why companies like OpenAI and AWS are currently competing for FDEs.

Since 2024, makandra has been implementing on-premises AI projects for companies operating in sensitive industries or where standard AI tools reach their limits. Part of our development team has since evolved into experts in on-premises AI applications. Or, as we might now say: into AI Forward Deployed Engineers.

We see this in our projects and in the project inquiries we receive: AI adoption in companies is moving forward, but more slowly than one might have thought three years ago, when the hype around significantly better language models began. The current trend toward FDE shows why this might be the case: standard AI tools are more often reaching their limits, but the necessary knowledge to build suitable AI applications in-house is frequently lacking.

The leap from prototype to production is greater for AI than for traditional software. A semantic search of a company’s own documents can be quickly built as a demo. Things get complicated when it involves sensitive data and every result must adhere to the company’s role- and permission-based system: Each user may only find what they are authorized to see. Added to this are integration with existing systems, quality assurance, and operations. An engineer who understands these challenges and builds solutions tailored to the client’s processes is well-equipped to handle this complexity.

There are three reasons to use Forward Deployed Engineering for AI projects:

• Complexity of the Subject Matter
  For business-critical, production-ready AI applications, data preparation, interfaces, quality assurance, and operations must all be considered and implemented. A project can therefore quickly become too large and complex for a team to handle alongside their daily tasks.
• Specialization of AI Experts
  Developing AI applications has evolved into a specialized field of its own, with models and best practices constantly changing. Internal teams are already fully occupied with developing existing systems and find it difficult to build up this necessary, additional expertise on the side. It may make more sense to bring in external expertise for the duration of the project.
• Responsibility Instead of Body Leasing
  A FDE takes responsibility for the system. He or she understands the business processes, designs the appropriate AI application, and integrates it via software interfaces to ensure the desired outcome is achieved. For this reason, simply augmenting the team through body leasing to implement the AI application is not the best solution.

This is the approach we take to implementing AI projects at makandra: We first help identify the right use case, design the AI application, and implement it in close collaboration with the client – often starting with a proof of concept, which we then develop into a production system. The operational application is handed over in a way that allows the team to work with and manage it.

Forward Deployed Engineering describes how, from our perspective, custom software development has always worked in complex projects. Especially when it comes to business-critical software, the provider acts in an advisory capacity from the very beginning, helping to understand the business processes and translate them into meaningful software. The software is tailored to the customer’s data requirements and tested in a live production environment before it is handed over to the customer.

We’ve been working according to this model for over 15 years – previously in software development, and now also in the development of local AI applications. What’s new about the current trend is the term itself and the fact that platform providers now want to work this way too, essentially entering the service provider business. This is not a criticism of this trend; on the contrary, it confirms what matters most in AI projects and can simultaneously allay the fears of young people interested in IT: Jobs in software development will not disappear, they will evolve. AI is transforming the tech industry, but at the same time, it offers many opportunities for employees and companies to grow, explore new fields, and help shape technological change.

The trend toward forward-deployed engineering in the AI field confirms a pattern that is repeated in AI projects everywhere: embedding AI within a company requires engineering work backed by a solid understanding of the client’s processes. Being able to understand business processes and find technical solutions: This has always been our approach for business-critical software projects, and it now benefits us in the development of local AI applications.

The fact that the world’s largest technology companies are now building dedicated organizations with budgets in the billions for this purpose shows that access to ever-improving language models alone is not enough to tackle the massive task of AI adoption. It also requires skilled engineers who can understand business domains and have the technical expertise to build AI applications for enterprises that deliver the much-promised relief through AI.