Salesforce Tripled Its FDE Bench to 1,000. The Feedstock Was Support Engineers.

If you opened a Forward Deployed Engineer req in Q2 2026 and started keyword-searching "FDE" on LinkedIn, you are losing. Salesforce just ran the largest natural experiment anyone has on this role, and the answer is not in the title field. It is in the support org down the hall.

In April 2026, Salesforce Ben reported that Salesforce's FDE team had crossed 1,000 people in March, tripling in six months. In the same fiscal year, Marc Benioff told TBPN that Salesforce hired zero new software engineers and quietly cut another roughly 1,000 roles starting in February across marketing, data analytics, and Agentforce. The 1,000-person FDE org was not built by recruiting. It was built by redeploying support engineers, success architects, and professional services staff into customer-embedded AI pods.

That is the playbook. And the same week Salesforce Ben published, Crunchbase logged 24,332 more US tech layoffs, including 400 at Workday's Global Customer Operations on February 4. The exact profile Salesforce just proved is the best FDE feedstock is now sitting on the market in volume.

The Salesforce A/B test, in plain numbers

Three datapoints from the Salesforce newsroom and Benioff's own interviews change how this req should be sourced:

9,000 → 5,000

Salesforce tripled the FDE team from three internal feeder groups: engineering, professional services, and customer success. Hiring managers inside Salesforce repeatedly told the newsroom that the redeployed support engineers were among the best hires they had ever made. That is not a PR line; it is a sourcing signal. The people who spent four years debugging Apex triggers for Fortune 500 admins already know the product, the integrations, the political map of an enterprise buyer, and how to deliver bad news without losing the account.

What they did not have, and what Salesforce trained in, was the Python plus RAG plus agent-orchestration layer. That gap is six weeks of focused work for a senior SE. It is not a hiring problem.

Why "Forward Deployed Engineer" search misses the pool

FDE postings spiked 800% on Indeed between January and September 2025. Average total comp is $238,000, staff-level clears $630,000, and New York now accounts for 35% of all FDE postings versus San Francisco at 11%. Every applied AI shop wants one. OpenAI, Anthropic, Cohere, Databricks, Ramp, Rippling, Intercom, and Adobe all run FDE teams now. Colin Jarvis grew OpenAI's FDE org from 2 to over 10 in a year. Leo Mehr took Ramp's from 2 to 16 in eighteen months.

The candidates with "Forward Deployed Engineer" already in their LinkedIn headline number in the low hundreds. Every recruiter in the Valley is messaging the same people. If your applied AI engineer pipeline depends on title match, you are competing with OpenAI's in-house team for a pool that fits in one conference room.

The actual pool is larger by an order of magnitude, and it is mislabeled. Senior Technical Support Engineers, Solutions Engineers, Customer Engineers, Implementation Architects, and Success Architects at Salesforce, Workday, ServiceNow, Zendesk, Okta, and the long tail of Salesforce ISVs. Their skills map cleanly onto the required FDE stack: Python, API integration, distributed systems literacy, RAG and vector store familiarity, one major cloud, and the ability to explain a deployment to a non-technical VP without losing the room.

The pod structure tells you what to source for

Salesforce builds its FDE pods as one deployment strategist plus two FDEs, each pod focused full-time on a single client for roughly three months. The strategist owns the AI strategy conversation; the two FDEs design, build, and ship the agent. That means two separate sourcing motions, not one generic req:

• Strategist: ex-consulting (Deloitte, Accenture, the FDE Partner Network firms like Appiphony, Aquiva Labs, Bridgenext) plus demonstrable AI literacy. Often a former Solution Architect or CTA.
• FDE proper: five-plus years at an enterprise SaaS company in a technical customer-facing role, with shipped Python in their GitHub or in case studies, comfort with at least one LLM API, and ideally a side project that touches RAG or agents.

Posting one JD for both gets you the wrong shortlist twice.

Where the supply is right now

As of May 15, 2026, there have been 179 layoff events this year affecting 113,863 workers, roughly 843 job losses per day. That tracker is the supply curve for your FDE candidate pool. Specifically:

• Workday GCO, February 4, 2026: 400 cut, support-heavy. Most have eight years of enterprise SaaS deployment scars.
• Salesforce support, ongoing: the difference between 9,000 and 5,000 is roughly 4,000 engineers. A meaningful share were redeployed internally. The rest are on the market.
• Salesforce-adjacent ISVs and the Crunchbase 2026 tracker: Cloudflare, Freshworks, ZoomInfo, Okta, PayPal, Upwork. Support and SE roles disproportionately represented.

These people are not posting "open to work" with "Forward Deployed Engineer" in their headline. They are posting with their current title, which is some variation of Technical Support Engineer III or Principal Solutions Engineer, and a skills section that mentions Salesforce, Apex, Python, REST, and increasingly LangChain or LlamaIndex.

This is exactly the search that breaks Boolean. You want people whose implicit profile matches an FDE, not people whose explicit title does. Which is why we built Refolk: you describe the candidate in plain English ("senior technical support or solutions engineer at Salesforce, Workday, or a Salesforce ISV, five-plus years, Python in their GitHub, bonus for any LLM or RAG project") and get back a ranked shortlist across LinkedIn, GitHub, and the open web. No title field gymnastics.

The people who spent four years debugging Apex for Fortune 500 admins already know how to deliver bad news without losing the account. That is the moat.

What to actually screen for

Coding is table stakes. The Salesforce newsroom is blunt about this: what separates strong FDEs from average ones is judgment, pattern recognition, and the ability to deliver hard truths to a C-suite stakeholder without alienating them. Those qualities do not show up cleanly on a résumé and they definitely do not show up on a LeetCode score.

Concrete screening signals that map to FDE success, in order of how much they actually predict performance:

1. Customer-facing tenure at a complex product. Five-plus years at one enterprise SaaS company is worth more than ten years across five startups. Depth beats breadth here.
2. Shipped integrations. Did they write the Workday-to-Salesforce sync? The custom Snowflake connector? Real artifacts beat job titles.
3. Internal escalation reputation. Hard to source for directly, but references will tell you in thirty seconds. The phrase you want to hear is "we sent her into the bad accounts."
4. Python or TypeScript fluency, not mastery. They will pair with an FDE lead for the first two deployments. You are hiring for trajectory, not for a senior IC who can architect from cold.
5. Any LLM-era side project. A weekend RAG demo on their GitHub is a stronger signal than three years on the Agentforce launch team without it. It tells you they pattern-match on the new stack on their own time.

Geographic reality

New York is now the FDE capital. 35% of postings, against San Francisco's 11%. But the supply is more distributed than the demand. Internal data on the support engineer pool we just described concentrates in NYC, Atlanta, and Salt Lake City, with secondary clusters in Indianapolis (Salesforce), Pleasanton (Workday alumni), and Raleigh.

If your FDE req is NYC-anchored and you are only sourcing in the five boroughs, you are missing two of the three densest pools. Remote-first reqs win this market. Hybrid reqs that demand three days in Manhattan lose to Anthropic and Ramp on comp anyway; the way to win is to fish in Salt Lake and Atlanta where the supply is thicker and the comp expectations are 20% lower.

The window is short

Senior Solution Architects and CTAs are moving but rarely available longer than four weeks. Q2 2026 is the peak availability window for redeploy support engineers AI shops want, because the February cuts are aging out of severance and the Big Four consultancies are about to absorb the strategist-shaped subset.

If you have an FDE req open right now and you have spent the last two weeks messaging people with "Forward Deployed Engineer" in their headline, stop. Rebuild the search around the profile, not the title. Salesforce already proved which one works.

A practical week-one plan:

• Pull the Workday February 4 cohort and the Salesforce support alumni who left in the trailing six months. Filter for Python signal.
• Pull current Solutions Engineers and Technical Support Engineers III+ at Salesforce ISVs (Hex, 1Password, Veriff, Envoy, NetDocuments). These people are not laid off but they are watching the redeploy story closely and they pick up the phone.
• Build a separate strategist list from ex-Deloitte AI practice, Accenture's Salesforce SBU, and the named PDOs (Appiphony, Aquiva Labs, Bridgenext).
• Skip anyone whose only AI signal is "AI enthusiast" in the headline. You want commit history or shipped product.

That is your FDE candidate pool. Refolk can compress the first three bullets into a single plain-English query and rank the output by fit. The fourth one is judgment, and that part is still yours.

Palantir invented this role in the early 2010s, called them Deltas, and until 2016 ran more Deltas than software engineers. The model is not new. What is new is the scale (1,000-plus at Salesforce alone) and the supply (every enterprise SaaS support org is shrinking at once). The companies that figure out the title-versus-skill mismatch in the next ninety days will staff their pods. The ones still sending InMails to people with "FDE" in their headline will pay $630,000 for staff-level talent and still come up short.

FAQ

Is the support engineer to FDE transition real, or is it a Salesforce PR narrative?

It is real, and Salesforce is not the only proof. Ramp, OpenAI, and Intercom have all promoted internal support and SE staff into FDE roles. The Salesforce newsroom quote from hiring managers calling the redeploys "some of the best hires they had ever made" is consistent with what FDE leads at smaller shops say privately: customer judgment is harder to teach than Python, and senior support engineers arrive with it pre-installed.

How do I screen for FDE potential when the candidate has zero LLM experience on their résumé?

Ask for any side project from the last twelve months, technical or otherwise. You are screening for self-directed pattern recognition, not for prior Anthropic experience. A senior SE who built a weekend Retool app to automate their own ticket triage is a stronger FDE candidate than one who took a Coursera RAG course. Then run a paired coding interview with an existing FDE; trajectory shows up inside ninety minutes.

What comp should I offer to pull a Workday or Salesforce support engineer into an FDE seat?

Average total FDE comp is $238,000 and staff-level clears $630,000, but those numbers skew toward NYC and toward people who already carry the title. A Salt Lake City Salesforce support engineer with eight years of tenure is currently making $145,000 to $180,000 all-in. A $220,000 to $260,000 FDE offer with meaningful equity is a 40% raise and reads as life-changing. You do not need to pay NYC FDE comp to win in tier-two metros.

Should I source the deployment strategist and the FDE on the same req?

No. The pod structure is one strategist plus two FDEs, and the strategist profile (ex-consulting plus AI literacy, often a former CTA or Solution Architect) is a separate sourcing motion from the FDE profile (ex-support or SE plus Python and RAG). Combining them produces a shortlist that fits neither role. Run two reqs, two pipelines, and accept that the strategist is harder and slower to fill.