How Many Companies Hire Forward Deployed Engineers?
Our registry tracks 111 companies across FDE and adjacent AI roles: 88 core FDE, 15 ML engineering and 8 applied AI. Coverage and evidence limits explained.
BY ARJUN MEHTA · FDEINTERVIEWS EDITORIAL · UPDATED OCTOBER 9, 2026 · 4 MIN READ
PRACTICE THIS:Every company we track, with its loop ·The must-know FDE questions ·Customer and behavioral rounds ·System design under customer constraints
We track 111 companies across forward deployed engineering and adjacent AI roles, as of October 9, 2026. Our registry classifies 88 as core FDE, with another 23 in ML engineering or applied AI.
This measures the companies we cover. It does not count current vacancies, completed hires or every employer in the market. A company stays useful for preparation after an individual job posting closes.
The breakdown
| Registry classification | Companies |
|---|---|
| Core FDE | 88 |
| ML engineering | 15 |
| Applied AI | 8 |
| Total tracked | 111 |
Every company has an interview-process entry, including entries that explicitly say the rounds are unknown. The company directory links to the role context and available evidence.
How much we know about the interviews
| Current evidence label | Companies |
|---|---|
| Documented | 13 |
| Partial | 36 |
| Limited | 62 |
These are editorial classifications, not employer endorsements. A partial entry may contain a general company hiring framework, a candidate's experience or a sequence whose applicability to FDE remains uncertain. A limited entry gives role context without inventing stages.
The September audit searched all 101 companies previously in the registry and added six: SAP, Figma, Atlassian, Accenture, Reducto and P-1 AI. It revised 27 existing entries. Searching a company did not automatically verify its entire process. Where only a role was checked, the source note says so and the older process review date remains unchanged.
What qualifies a company for inclusion
We look for employer postings or other attributable evidence of the relevant work. A role description can establish customer deployment responsibilities; it cannot establish interview stages by itself.
The classification also matters. Some readers want an explicit FDE title. Others are considering applied AI, solutions or ML engineering roles with overlapping preparation. Inclusion in one registry does not mean all these jobs have the same customer responsibilities or interview loop.
Start with the exact role and location on the employer's posting. Then check the scope above the interview stages. A general engineering guide may be useful while leaving the FDE-specific assignment unresolved.
What changed in the reviews
An October 5 sweep, integrated on October 9, searched all 107 companies then in the registry and added Goodfire, with role evidence but no interview stages. It also revised six existing entries. Deployment Inc now documents its public hiring-challenge route, while xAI and McKinsey QuantumBlack moved from documented to partial because their older FDE-specific detail could not be traced to a source for that role. Search coverage does not mean every legacy claim was reverified.
The October 7 targeted refresh added Retell AI, Coder and Render. Retell publishes a senior-role sequence; the other two entries retain explicit limits on known rounds. This was a targeted refresh, not a new verification of all 110 employers. Current openings appear separately on the FDE job board.
In September, role naming needed correction as well as new additions. AWS, Baidu, Cohere and Databricks have explicit forward deployed roles or organizations reflected in the refreshed classifications.
Several processes were also downgraded from documented to partial after review. A detailed list is not necessarily a verified list. We kept useful preparation guidance, preserved the previous records for review, and separated employer guidance from individual candidate experiences.
How to use and cite the count
Use the directory to build a target list, then verify the current opening and its requirements. Compensation and vacancies change independently of interview guidance; neither should be inferred from a fresh source-check date.
If citing the count, attribute it as 111 companies tracked by FDEInterviews.com as of October 9, 2026. Do not describe it as the total number hiring worldwide.
Method
Counts come from the company classifications and confidence labels in our registry. The build checks the published headline figures against the underlying data. Sources on individual pages distinguish company guidance, candidate reports and reporting, with separate publication and review dates when known. This audit did not comprehensively reverify legacy compensation figures.
Turn it into offers. Work the real questions and concepts this maps to:
FAQ
FDEInterviews.com tracks 111 companies as of October 9, 2026: 88 classified as core FDE, 15 as ML engineering and 8 as applied AI. This is a research registry, not a count of currently open vacancies or a worldwide hiring census.
No. It counts our registry, assembled from employer postings, company guidance and public reports. A company can remain in the registry after a posting closes. Check the linked employer page for current availability.
We have not measured hiring volume across companies. Palantir, OpenAI, Anthropic, Google, AWS and other employers have relevant roles, but this registry does not support ranking them by the number of engineers hired.
Growth in our registry measures research coverage. It does not establish a market growth rate. New employer postings can show that a role exists without showing how many people will be hired.
Discussion (5)
The core against adjacent split is the useful part here. I have interviewed at three places that advertised machine learning engineer and the loop was pure FDE: a customer scenario, a messy data exercise, and a stakeholder conversation. If you filter job boards on the exact title you will miss most of the market.
That is exactly why the registry carries both. Filtering on the literal string is the single most common way candidates undercount their options. The tell is in the job description rather than the title: if it mentions customers, deployments or embedding with a client team, the loop will look forward deployed whatever the header says.
Reader question: does a company appearing in the registry mean they are hiring right now, or that they have hired for this at some point?
At some point, and that is a deliberate choice. Postings open and close weekly, so a live-vacancy list would be stale within days and would tell you nothing about how to prepare. The registry answers a slower question, which is where this role exists and what the loop looks like when it runs.
Worth saying out loud that the 62 companies with limited process detail are not a gap in the research so much as a reflection of reality. Most companies simply do not publish their loop. Treating absence of detail as evidence of an easy loop would be a mistake.