The role that sells, builds and survives the promises a platform company makes.
Twelve months ago this was a Palantir curiosity. As of June 2026 it is a structural layer of the AI economy — capitalized with billions, hired by the hundreds, and copied by every consultancy with a logo.
— Field report, June 2026
Not a job title. A go-to-market strategy.
For products that only win through deep, technical integration.
Embed with the customer. Turn a vague, painful problem into a concrete technical plan. Ship the code that makes the product real inside their environment — then hand it over, working.
Between the product org and the field. Accountable to customer outcomes, paid by the platform company — a deliberate tension that keeps the roadmap honest.
Modern platforms don’t ship in a box. They land in fragmented data, legacy systems and organizations mid-transformation. Someone has to make the landing soft.
From Palantir’s secret weapon to the industry’s playbook.
The substrate that births the role: government data, sensitive environments, nothing off the shelf.
Palantir embeds engineers directly with customers — internally, “Deltas.” Through 2016 they outnumber core software engineers.
Field lessons harden into product. Deltas cycle back to core engineering, carrying customer reality with them.
OpenAI stands up Solutions Architects: MVPs on anonymized data, building the case for a deeper role.
OpenAI, Anthropic, Stripe and Ramp stand up formal FDE orgs. Salesforce commits to a team of 1,000. Postings grow +1,165% YoY.
Deloitte, Accenture, KPMG and BCG collectively out-post every product company except Google. EY launches a dedicated FDE practice.
OpenAI launches the Deployment Company — $4B+ raised, Tomoro acquired, ~$10B valuation. Days earlier, Anthropic unveils a $1.5B enterprise-services venture with Blackstone, Hellman & Friedman and Goldman Sachs.
Google Cloud begins hiring hundreds of FDEs to push frontier models into Fortune 500 workflows. “The era of the pilot is over.”
FDEs own the glue layer between raw models and production: retrieval, tools, evals, security — the last mile that makes AI real.
Palantir turned services into a moat. The industry spent 2025 copying the trick — and 2026 capitalizing it.
— Field report, June 2026
Not a solutions architect. Not a sales engineer.
The heuristic: if it doesn’t write production code on the customer’s infrastructure, it isn’t an FDE — whatever the job title says.
Numbers are typical — postings vary wildly. (bloomberry · FDE Pulse, 2026)
Six things the job actually is.
Scoping the real problem
Walk into a vague brief — “cut fraud,” “use AI somewhere” — and leave with a concrete technical plan that maps customer pain to systems.
Prototyping fast
A working proof inside the customer’s data in days, not quarters. Code, not slideware.
Hardening to production
Rewrite the prototype to survive scale, security review and someone else’s infrastructure. Where most vendors bail; FDEs don’t.
Integration & data
Wire into fragmented stacks: legacy systems, multi-cloud, proprietary schemas. Move data. Design pipelines. Untangle identity.
Technical ownership
Own the lifecycle. Train the customer’s engineers, then hand over when it’s safer in their hands than yours.
Field → roadmap
Carry patterns and edge cases back to the product team. The best feature ideas never come from a planning doc.
A day, observed.
A composite Monday from four FDEs across Palantir, OpenAI and a Series-D fintech, mid-rollout.
Check the customer-side on-call channel. A batch job failed overnight; diagnose before standup.
Sync with the customer’s data team. You run this — without you, there is no cadence.
Whiteboard the retrieval layer with two of the customer’s senior engineers. Disagree productively.
Casual time with the VP who sponsors the project. This is the actual account strategy.
Two PRs against the customer’s repo. Tests pass. Deploy to staging behind a flag.
Thirty minutes with core product — a missing primitive is blocking three accounts. You have the spec ready.
Draft the internal case study. Half engineering post-mortem, half field report.
Three disciplines. One operator.
Select a discipline to inspect it.
Engineering depth
You ship production-grade code on infrastructure you don’t own. Not prototype-grade. Not configuration. Code that runs at 3 a.m. when nobody is watching.
Where the market is going.
Twelve months ago this was a hiring trend. As of June 2026 it is a structural layer of the AI economy — capitalized, consolidated, and specializing fast.
YoY growth in postings — 643 → 5,300+ (Indeed, Apr ’25 → Apr ’26)
Committed to FDE ventures in one month — OpenAI ($4B+) and Anthropic ($1.5B), May ’26
Of enterprise AI pilots fail to reach P&L impact — failures trace to deployment, not models (MIT)
Of postings now in New York — surpassing San Francisco’s 11% as the FDE capital
The deployment-company era
In a single fortnight, FDE pods graduated from team to asset class. OpenAI launched the Deployment Company — majority-owned, $4B+ raised from TPG, Advent, Bain and Brookfield, led by COO Brad Lightcap, with Tomoro’s ~150 deployment engineers acquired on day one. Days earlier, Anthropic unveiled a $1.5B enterprise-services venture with Blackstone, Hellman & Friedman and Goldman Sachs — pitched as democratizing access to forward-deployed engineers for the mid-market. Engineering-as-outcome is now a fundable, standalone business.
Consulting retools
EY launched a dedicated FDE practice. Deloitte, Accenture, KPMG and BCG collectively post more FDE roles than any product company except Google. McKinsey’s QuantumBlack now demands 8+ years of real engineering. The role has migrated from AI-pure companies to anyone deploying someone else’s platform.
Hyperscalers go forward
Google Cloud is hiring FDEs by the hundreds — “the era of the pilot is over.” Microsoft and AWS are competing for the same talent. The labs offer frontier access and equity; the clouds offer distribution and a Fortune 500 Rolodex. Both sides will keep raiding each other through 2026.
The job specializes
Eval engineering is the 2026 non-negotiable — suites that catch hallucinations and regressions before production. Agent operations, retrieval engineering and sovereign-AI deployment are emerging as named subspecialties with their own bands.
Down-market and distributed
PE-portfolio programs push FDEs beyond the Fortune 500 into the mid-market. 59% of FDE-hiring startups are Seed through Series A. A third of roles are remote-eligible — but presence is still the product: most postings remain on-site or hybrid.
Who is betting on the role.
The original — more open FDE roles than the next two companies combined.
The Deployment Company: $4B+ raised, Tomoro acquired, led by Brad Lightcap.
Enterprise-services venture with Blackstone, Hellman & Friedman and Goldman Sachs.
“The era of the pilot is over.” Embedded builders for Fortune 500 AI workflows.
Momentum program; highest-stakes agent deployments.
Lakehouse and agent deployments inside enterprise data estates.
Sovereign-AI deployments across European enterprise and government.
Enterprise LLM deployments and fine-tuning.
Consulting’s biggest FDE bet — deploying third-party platforms at scale.
Dedicated FDE roles launched in UK & Ireland — governance from day one.
8+ years of real engineering required — consulting collides with code.
Revenue & Financial Automation — closing PMF gaps in the field.
Enterprise migrations; custom integrations at the ledger.
Forward Deployed AI Engineers building on Firefly — the giants adopt the model.
Government and enterprise AI programs.
Native apps and AI workloads inside customer accounts.
— Partial list. Open-role counts from live job-board datasets, May–June 2026.
How the market prices an FDE.
Tick = median. Band = typical 10th–90th percentile. Compiled from published bands at Palantir, OpenAI, Anthropic and Google Cloud, plus FDE Pulse and Levels data, 2026. The market is bifurcated — posted packages run $53K to $1.2M.
YoY growth in FDE postings — the fastest-growing title in tech against a flat hiring market. (Indeed)
Where lab packages have stabilized for mid-to-senior FDEs — benchmarked against top researchers.
What Staff-level FDEs now clear at Palantir. Premiums follow proven deployment-to-renewal impact.
FDE postings carrying a revenue quota. It is an engineering role; with a quota it is something else.