Forward Deployed Engineer

What is a Forward Deployed Engineer?

A forward deployed engineer (FDE) is a software engineer who works embedded with a customer instead of on a product team at headquarters. The FDE learns how the customer's work actually gets done, then writes the code, integrations and, increasingly, AI agents that change it, and stays until the results are measured.

Palantir popularized the role, sending engineers into government and enterprise customers to make its software work on their real data. In 2025 and 2026, AI labs such as OpenAI and Anthropic, and agent-deployment companies such as Varick Agents, built large FDE teams, because the hard part of enterprise AI turned out to be fitting models into messy processes, not the models themselves.

In AI work, the FDE typically maps a department's processes, decides which steps should be deleted, hard-coded, handed to an agent or kept with a human, builds agents inside the systems the company already uses (Salesforce, NetSuite, Dynamics, Workday), and reports against a baseline months later.

Key Characteristics

• Process mapping first: Interviews with the people who do the work, process mining over the systems of record, and a review of existing documentation, because the real process is rarely written down.
• Step-by-step sorting: Each step is deleted, replaced with deterministic code, given to an agent where judgment and historical data exist, or kept with a human in the loop for risky actions like payments.
• Building inside existing systems: Agents act in the CRM or ERP the client already runs, with approvals in tools like Slack, so staff don't have to learn a new surface.
• Three skill sets in one: Business process knowledge, production engineering with auditability and security, and the AI layer (model selection, evals, harness tuning and rollbacks).
• Measured outcomes: Cycle time, cost per transaction, straight-through rates and accuracy, measured before the build and again at three and six months.

Why Forward Deployed Engineers Matter

Most AI pilots fail because AI gets layered on top of processes nobody has mapped. FDEs fill that gap. In one anonymized accounts payable engagement described by Varick Agents CEO Vasuman Moza, mapping and rebuilding the process took it from 17 steps to seven, cut cycle time from 24 days to six, raised the share of invoices processed without exceptions from 18% to 87%, and cut cost per invoice from $31 to $6.

That value is why the role pays well. On Greg Isenberg's channel the argument is simple: an engineer who drives $10 million of value can earn 10% of it, and some private equity firms give FDEs a share of the carry on companies they transform. The same skills apply inside a company that wants to build its own internal FDE team.

Historical Context

The term comes from Palantir, where "forward deployed" engineers worked on-site with customers in contrast to engineers building the core product. The AI agent wave of 2025 and 2026 turned it from a niche title into one of the most in-demand engineering roles, as AI labs, consultancies, PE-backed AI roll-ups and agent startups all competed to hire people who can both understand a business process and ship production code.

Related Reading

• Vasuman Moza: How Varick's Forward Deployed Engineers Work - Concrete FDE engagements and numbers
• AI Agents - What FDEs deploy
• Workflow Automation - The broader practice
• Enterprise AI - Where FDEs most often work