Forward Deployed Engineer: The New Hottest Job in the AI Era – What Exactly Do They Do?

An AI Job “Arms Race”

Let’s start with some recent news about a new AI role: Forward Deployed Engineer (FDE).

Google is doubling down on FDE hires and has dramatically simplified the interview process. Thomas Kurian, CEO of Google Cloud, announced the creation of a brand new, AI-focused unit under the Go-To-Market team, and they are aggressively recruiting FDEs for it.

Word is that their interview process has been compressed from weeks and 4-6 rounds down to just two rounds over two days. It’s clear Google is not just eager but almost impatient to fill these positions.

On Monday (May 11), OpenAI announced the formation of “The OpenAI Deployment Company.” This is an independent entity funded by a $4 billion private equity investment, valued at $14 billion, with investors including TPG, Advent, and others. OpenAI itself doesn’t seem to be a direct investor but rather a partner.

The announcement specifically mentioned FDEs, stating their role is to “work closely with business leaders, operations teams, and frontline staff to precisely identify areas where AI can generate the most value, redesign the organization’s infrastructure and key workflows around AI, and ultimately convert these benefits into persistent, stable systems.”

Thus, FDEs will play a critical role in OpenAI’s enterprise sales business. Their mission is to ensure the company’s AI systems work in real-world customer operations and genuinely create value. Outsourcing this function to the newly formed “Deployment Company” also allows OpenAI to focus on developing more powerful AI models, letting partners and their FDEs handle the messy customer integration work.

A related development: OpenAI acquired Tomoro. This is a UK-based AI company founded in 2023, with 150 FDEs across the UK, Asia, and Australia. This was the first acquisition since the “OpenAI Deployment Company” was established.

Anthropic is following the same playbook, creating its own independent FDE consulting firm. On Monday (May 4), Anthropic published an extremely vague announcement about this new business, without even revealing its name and with few investment details.

Known investors include Anthropic itself, Blackstone, Hellman & Friedman, and Goldman Sachs. The new company’s mission is to “partner with mid-sized enterprises across various industries to bring the large language model (LLM) Claude into their most important business operations.”

Anthropic’s strategy seems identical to OpenAI’s: use external capital to form an independent company, where FDEs help enterprises integrate Claude into their systems. It’s foreseeable that this will drive the number of Claude tokens purchased by these enterprises to record highs.

Explaining FDE in Plain English

So what exactly is an FDE? The full name is Forward Deployed Engineer, or FDE for short. The literal translation is “frontline deployment engineer,” but the name alone doesn’t clearly explain what they do.

One-sentence version: An engineer who sits on-site at the client company and writes code.

In more detail, this role sits somewhere between a software engineer, a solutions architect, and a consultant, but is more hands-on. They sit directly at the client’s company and use their company’s AI technology to solve the client’s real problems.

You might ask, isn’t this just a consultant? Not quite. Consultants typically give you a PowerPoint deck and tell you “what’s best to do.” An FDE gives you the code to make it happen. A solutions architect usually draws architecture diagrams and writes technical proposals. An FDE does that too, but also gets their hands dirty writing code, tweaking APIs, and debugging issues on-site.

If we had to put rough percentages on it: 25% writing code, 50% integration and debugging, 25% meetings and communication. In reality, the quiet time for pure coding is probably even less.

Actually, Palantir Is the Pioneer

Speaking of FDEs, this isn’t something new that emerged in the AI era. Palantir perfected this approach back in the 2010s.

Palantir built a data analytics platform, initially serving only the US military and intelligence agencies. Customer requirements were classified, making conventional communication impossible. So Palantir simply stationed engineers at client sites permanently. They observed customer needs up close and iterated rapidly on the spot.

These on-site engineers (Palantir called them “Deltas”) did more than just deliver projects. Their even more important task was: extract common requirements from the client side and feed them back to the product team to build as standardized features.

By 2016, Palantir had more FDEs than regular engineers, truly defining the role.

Three Companies, Three Different Paths for FDE

OpenAI is the most aggressive. They formed the OpenAI Deployment Company, bringing in TPG, McKinsey, Bain, and Capgemini. They even valued it at $14 billion and immediately bought a UK company with 150 FDEs ready to go. A promised minimum 17.5% return rate makes it look more like investing in infrastructure.

Anthropic is more measured. They found Wall Street giants like Blackstone, Goldman Sachs, and Apollo to form a joint venture. With an initial $1.5 billion investment, they are targeting the mid-size enterprise market. These investors hold a huge portfolio of companies, which are naturally the best user base for the Claude model.

Google is the most traditional. They hire their own people, with FDE positions distributed globally and decent salaries – senior roles in the US can have total compensation exceeding $400k. But the biggest difference is that Google’s FDEs get Google stock, while FDEs at OpenAI and Anthropic belong to independent companies, with no direct stake in the parent company.

Translating the “Real Talk” Behind Google’s FDE Job Posting

Job postings can be cryptic. Let’s translate one:

Even if it sounds a bit cynical, every company’s JD is similar. Having this mental preparation helps you figure out if the role suits you.

The Big Question: Is FDE Still Consulting?

Let’s look at three dimensions.

1. Organizational affiliation. Palantir’s FDEs are part of the product team, sharing the fate of the parent company. But FDEs at OpenAI and Anthropic belong to independent companies. This dilutes information flow, identity, and career progression.
2. Feedback loop. The biggest value of an FDE is discovering client needs and feeding them back to the product. But with an organizational gap between the independent company and the parent company, this feedback channel can be blocked. The FDE risks becoming a mere “code-writing consultant.”
3. Interest alignment. Google’s FDEs get parent company stock – interests are aligned. FDEs at OpenAI and Anthropic get a share of the independent company’s profits. Even if the parent company’s valuation goes through the roof, you don’t get a slice.

The conclusion: The FDE role at OpenAI and Anthropic has become closer to consulting, while Google’s is closer to the traditional FDE model.

Who Should Care About FDE?

Let’s break it down by three groups:

• New graduates: An excellent opportunity. Software engineering roles at big tech are shrinking, but FDE is hiring heavily. You get quick exposure to enterprise AI projects and faster growth.
• Senior engineers: You might feel it’s a “downgrade”. Clients change frequently, lacking long-term belonging. But if you’re thinking about starting a company or want to get closer to the business side, FDE is a fantastic window into deep enterprise needs.
• Non-technical people: The barrier is still high. This isn’t something you can pick up after a few months of Python.

The AI Race Has Quietly Shifted

For the past three years, the AI industry has been competing on model size and benchmark scores. Now the question has changed – most enterprises don’t lack models; they lack someone to help them integrate the model into their business.

OpenAI put in $4 billion. Anthropic invested $1.5 billion. Google compressed hiring to two days. These massive investments show that AI companies are changing how they make money – from selling models to selling deployment.

On a larger scale, for every $1 spent training a model, you might need to spend another $1 to actually make it run.

FDE stands right at the forefront of this inflection point.