The number
Hundreds. That's how many Forward Deployed Engineers Google Cloud is now hiring, per CEO Thomas Kurian — an in-house deployment corps embedded inside customer offices to ship production AI. The disclosed bands run $127K–$183K base for the New York and Atlanta postings, with senior total comp reported as high as ~$700K. The timing is the story: one week after OpenAI spun its deployment arm into a $4B outside company, Google is scaling the same motion internally. The two biggest AI platforms are now competing on deployment headcount, not just benchmark scores — the fastest tell yet that shipping the model has become harder than training it.
What changed this week
The Plank census is still frozen — observed: 2026-07-04, count 1,206, now six-plus weeks static — so the signal is composition, not headcount. The clearest move in this week's curated reqs: the FDE role is going international.
Five of the seats added this week are outside the US. AssemblyAI opened a Senior FDE in London (£100K–£150K OTE). Cartesia planted its real-time-voice FDE role in Bangalore — the same seat it runs in SF and London, now in India's enterprise market. Edra and Diligent both posted London FDEs; OutSystems put a Principal Agentic-AI FDE in the UK; and Deployment Inc — a firm whose entire name is the thesis — is building a founding FDE team in Gurugram. London and Bangalore are quietly becoming the role's two offshore hubs, tracking where enterprise AI budgets and English-language delivery talent overlap. For US candidates the read is narrower comp arbitrage over time; for the market, it's confirmation the title has cleared "US-startup fad" and become a global delivery function.
New roles
CoreWeave — Forward Deployed Engineer, AI Agents San Francisco / Sunnyvale, CA · $182K–$242K base + equity A rare public-company (Nasdaq: CRWV) FDE band, and the board's first W&B Weave-observability seat since CoreWeave's Weights & Biases acquisition. Embed inside customers' AI teams and ship agents on the Weave toolkit. View role →
August — Forward Deployed Engineer New York City (Onsite) · $170K–$300K A $7M seed shop paying near-frontier money to put an FDE onsite with law firms. NEA-backed legal AI betting that white-glove deployment — not model quality — wins the category. All-onsite, no remote hedge. View role →
Plain — Forward Deployed Engineer San Francisco (Hybrid) · $180K–$250K + equity Plain's first FDE — the classic tell that an API-first startup's customers now need custom integration work, not just a signup flow. Ground-floor deployment seat with SF customers. View role →
Picogrid — Forward Deployed Engineer El Segundo, CA or Fort Bragg, NC · $120K–$200K base + options Defense-tech FDE working shoulder-to-shoulder with military customers on TAK/Maven/Lattice and multi-sensor fusion. Clearance preferred. A Fort Bragg seat is about as literally forward-deployed as the title gets. View role →
PhysicsX — Forward Deployed Software Engineer New York, NY · $180K–$250K base + equity Series-C AI-for-physics platform going into aerospace, defense, semiconductor, and automotive customers. FDSEs travel to sites 3–4 weeks a quarter across four continents. A rare seat that rewards deep simulation/ML over SaaS integration. View role →
AssemblyAI — Senior Forward Deployed Engineer London, England · £100K–£150K OTE + equity One of the leanest AI companies anywhere — under 100 people, ~$500K ARR per head, 600M+ monthly inference calls. That revenue density means unusual ownership per FDE. Series-C equity on top. View role →
Adonis — Senior Forward Deployed Engineer New York, NY (Hybrid) · $190K–$210K + equity Series-B ($31M) healthcare-payments AI. Lead on-site discovery at major health systems and write production code against Epic and Athena EHRs. An FDE that lives inside hospital revenue-cycle guts. View role →
One insight: only a third of the job is code
Perspective AI surveyed 1,500 Forward Deployed Engineers across 154 companies — Anthropic, OpenAI, Palantir, Scale, Databricks, Cohere, Sierra, Harvey, Glean, Mistral, and 140+ growth-stage shops. The headline finding isn't comp. It's the time split.
The median FDE spends 47% of the week customer-facing — discovery interviews, on-site deployments, design reviews — versus 31% actually shipping code and 22% on internal coordination. Read that again: the highest-paid engineering title of 2026 writes production code less than a third of the time.
That single ratio explains most of what's confusing about the role. It explains the comp premium — you're paying for a scarce blend of "can code" and "can sit across from a skeptical VP and own an outcome," and almost nobody has both. It explains the burnout risk the same survey flags: half your week is other people's calendars, in other people's buildings. And it explains why the internal-FDE seats we covered last week are spreading so fast — companies figured out that the 47% translates cleanly to their own business units, no airport required.
The practical read for candidates: interview for the 47%. Everyone preps the coding round. The FDEs who get hired — and who don't flame out — are the ones who can show a deployment they drove from a messy first customer meeting to a system running in production. The code is table stakes. The other two-thirds is the job.
One ask
Forward this to one engineer who's on the fence about an FDE move. The best thing you can do for someone eyeing the role is send them the week's real numbers — and it's the best thing you can do for this list, too.