Glossary & Sources
Every statistic in this course traces to an entry below, with a confidence rating. Where independent estimates disagree, the course shows the spread rather than the most flattering figure.
Glossary
Methodology
How the company scores were built
The seven-dimension scores in Module 07 are my reading of public job postings and company statements as at August 2026. They are not published by the companies, and no company has reviewed or endorsed them. They are written down in this detail specifically so that you can disagree with a particular score for a particular reason, rather than accepting or rejecting the whole thing.
Each company is scored 0–5 on seven dimensions. The first four combine into a Substance Score out of 100, weighted as shown. The remaining three are reported separately because compensation, accessibility and liveability are trade-offs you make — not measures of whether the role is genuinely forward deployed engineering.
| Dimension | Weight in Substance | What a 5 looks like | What a 0–1 looks like |
|---|---|---|---|
| Code Ownership | 30% | Owns production repositories, makes architecture decisions, reviews code | Configuration, demonstrations or ticket escalation only |
| Product Feedback | 25% | A named pathway from field work into core product, SDK or platform | Customer code accumulates in branches with no product route |
| End-to-End Scope | 25% | Discovery through architecture, build, production and adoption | One narrow slice of the lifecycle |
| Engineering Bar | 20% | Genuine production software-engineering assessment and expectation | Little or no engineering assessment |
| Compensation | — | Top of the global band for the level | Substantially below market for equivalent engineering work |
| Entry Accessibility | — | Genuine graduate and early-career entry points exist | Senior-by-default with a hard multi-year floor |
| Sustainability | — | Low, quantified travel and few structural constraints | Very high travel, clearance requirements or permanent on-site work |
Confidence ratings
Compensation assumptions
Chart design
Charts are hand-rolled inline SVG with no charting library. The categorical palette was validated
programmatically for colour-vision deficiency separation, lightness banding, chroma and contrast against
both the light (#ffffff) and dark (#161a20) surfaces this site actually renders on.
Every figure carries a table view, so no chart depends on colour perception alone. Scatter plots are limited
to three colour groups, which is the number that passes all-pairs separation in both themes.
Sources
Every entry states what it supports and what its limitations are. Company cards in Module 07 link directly to the relevant entries here.
Known limits of this course
Stated plainly, because a course that only tells you what it knows is less useful than one that also tells you where it stops.
Postings are not people
Every market figure here measures visible recruitment activity. There is no occupational code for FDE and no labour-force survey. Nobody knows how many people hold the role.
Some salary bands are estimates, though every posting is real
Every company in this course has a publicly verifiable Forward Deployed Engineer posting — each one is linked in the source index. But not every posting discloses pay. Where it does not, the band shown is constructed from comparable roles and the company card says so explicitly. Bands are disclosed for Palantir, OpenAI, Anthropic, Scale AI, Databricks, Google Cloud, Salesforce, Dexterity and Accenture; estimated for ServiceNow/Moveworks, Thought Machine, Meesho, Lumafield, Trase and TCS.
The growth figure is contested
The 729% headline comes from an indexed series. Job aggregators generate false positives when “forward”, “deployed” and “engineer” are matched separately. The defensible claim is that recruitment grew several-fold over 2025–26 — not a precise percentage.
Case-study impact is not attributable to individuals
Public case studies rarely disclose which employees held the FDE title. Reported outcomes reflect whole customer and product organisations, not one engineer.
Non-US data is thinner
US confidence is highest. Continental Europe, India and APAC rest on fewer disclosed ranges; the Middle East section is built on a handful of confirmed postings and is labelled Low throughout.
This is a snapshot
Data is current to August 2026. In a market growing this fast, treat anything more than six months old as directional. Job postings referenced here may have closed.