The Finance Engineer Role: Job Posting Statistics (2026)

Last updated: October 2026. Corpus frozen 26 October 2026. Second quarterly run against the frame first frozen on 27 July 2026.

"Finance engineer" is roughly a year old as a job title and has no research literature behind it: no AFP or APQC benchmark, no government occupational code, no salary survey. The numbers circulating about it come almost entirely from social posts and vendor blogs that cite each other, and none of them publish a denominator.

This page reports a first-party count instead. We queried the public board APIs of 1,061 candidate employers across the US and Europe, scanned every open posting on the 418 live boards that returned, filtered on the department each employer assigns to the role itself, adjudicated the matches against written rules, and coded all 27 surviving requisitions from full posting text. The frame is the same one used in July, unchanged, so the two runs can be compared directly.

At a glance

What changed since July

This is the first comparison the page can make at a frozen frame, and the reason the frame is frozen: any change below comes from the market, not from us adding employers.

The count fell, from 37 requisitions to 27, while finance hiring at the same employers rose 3%. Finance-department postings went from 987 to 1,018 on the same 1,061 identifiers, so the share of finance openings that are engineering roles moved from 2.6% to 2.0%. At counts this small, a ten-requisition swing is roughly 1.3 standard deviations: it does not establish a decline, and one more quarter is needed before calling a direction. What it does rule out is the "exploding role" narrative. On a fixed set of employers, the population did not grow.

The population almost entirely turned over. 32 of the 37 July requisitions are gone. Matched by ATS job ID, only five July requisitions are still open: Brex, Helsing, Harvey, and two that were retitled rather than closed. Coinbase renamed "Senior Staff Software Engineer, Finance Automation" to "Forward Deployed Engineer, Finance [Office of the CTO]" under the same job ID, and Stripe renamed "Software Engineer, Internal Systems" to "Software Engineer, Financial Business Insights". Klaviyo's Order to Cash Track Lead was reposted under a new ID. Twenty-two of the 27 October requisitions did not exist in July. A corpus of live postings is a snapshot of a moving population, and this one moves faster than the July data suggested.

Agent and LLM references went from 57% of postings to 70%. 19 of the 27 coded postings reference AI agents, agentic workflows or LLMs. The share of postings naming AI coding tools as a skill (Claude Code, Codex, Cursor) doubled, from 2 to 4. Model Context Protocol is named in 5 postings at 4 employers, against 4 at 3 in July, and none of July's three MCP employers still has a qualifying role open: the requirement was renewed entirely by new employers.

Pay held exactly. The median band midpoint across disclosed US ranges is $231,500 in both runs, on 17 ranges in July and 19 in October, with almost no requisition in common. The ceiling rose to $405,000 (two Anthropic Finance Engineering roles).

Method

Inclusion. A posting qualifies if, at the freeze date, it was live on an employer's own applicant tracking system and both:

1. it serves the employer's own finance function, not a finance-industry vertical and not a fintech product team, and
2. it explicitly requires building systems, automation, data pipelines or AI tooling, rather than merely using them.

Why we scanned departments, not titles. Employers do not agree on a title for this role: the 27 requisitions below carry 27 different titles, and 6 of them contain no finance word at all (16 of 37 in July). Title search cannot find this population. Instead we pulled every open posting from each board and filtered on the department the employer itself assigns in a structured field, which is what makes the reporting-line findings on this page measurable rather than inferred.

The endpoints are public and the query is reproducible:

• Greenhouse: boards-api.greenhouse.io/v1/boards/{token}/departments, then /jobs/{id} for full content
• Ashby: api.ashbyhq.com/posting-api/job-board/{slug}?includeCompensation=true

Sampling frame. 1,061 candidate board identifiers, assembled in July in two passes: first US and global technology, SaaS, fintech and AI employers, then a dedicated European pass covering scaleups and fintechs across the UK, Germany, France, the Nordics, the Netherlands, Iberia, Italy, Poland, Austria and Switzerland. The list has not been edited since, by rule: adding employers between runs would make the counts incomparable. 418 identifiers returned a live board with at least one open posting in October (421 in July). This is a convenience frame, not a random sample, and the Limitations section states what that costs.

Matching. Any title containing an engineering noun (engineer, engineering, architect, developer, tech lead) combined with a finance noun (finance, financial, FP&A, accounting, controller, treasury), plus any engineering-titled role placed by its employer in a Finance, Accounting, Finance Systems or Treasury department. That returned 97 raw matches, the same number as in July by coincidence.

Adjudication. The raw matches were then filtered against written exclusion rules, applied by hand and recorded:

R3 covers Harvey's three forward deployed roles, which are deployed into clients; R4 covers OpenAI's government support systems role, filed in its Finance department. R5 is new this run. It was added as its own rule rather than by widening an existing one, so the July exclusion counts stay exactly as published. That removed 67 of 97. The remaining 30 postings collapse to 27 distinct requisitions once roles posted across multiple locations are merged.

Coding. All 27 requisitions were retrieved in full text and coded for employer-declared department, stated reporting line, minimum years of experience, tools and platforms named, references to agents or LLMs, AI coding tools, disclosed pay, and, new this run, whether the posting requires evidence of past work (a portfolio, examples of automations shipped, links to projects). A field is recorded only where the posting states it. Where a posting states several experience thresholds, the lowest is recorded; a ceiling ("less than 2 years", in an internship) is not recorded as a minimum.

Archival. Job postings are taken down within months, and this run shows how fast: 32 of the 37 July links no longer lead to an open role. Two layers guard against that:

1. Full-text snapshots. All 27 postings were captured in full at the freeze date, each stored with its retrieval timestamp, source endpoint and a SHA-256 of the text, so any claim here can be checked against the exact wording we coded. The July snapshots are kept unchanged alongside them.
2. Independent third-party captures. Each live URL was submitted to the Internet Archive's Wayback Machine. 25 of the 27 are archived and linked in the appendix. The two exceptions are Klaviyo's and Helsing's roles, whose careers sites and Greenhouse mirrors still refuse to serve the archiver, as they did in July; those are marked "local only" and rest on layer 1 alone.

Two caveats on layer 2. Several of these boards (Ashby postings, and career pages at Stripe, Coinbase and Block) are JavaScript applications, so the archived page may show a shell rather than the posting text. And where a Wayback timestamp predates our freeze date, an existing capture was reused and may differ from the version we coded. In both cases the layer 1 snapshot is the authoritative record of what was counted, which is precisely why it exists.

The data. The full coded corpus is published as an open dataset, so nothing on this page has to be taken on trust:

layerz/finance-engineer-postings on Hugging Face, CC BY 4.0, one row per requisition per run.

Each row carries the employer-declared department, the title and location as posted, the ATS publication date, minimum years of experience, the tools named, flags for agent and MCP references, the pay band verbatim and unparsed, the reporting line, the live URL and the Wayback capture. A blank means the posting did not state it, never that the value is zero.

Download it, recount it, and disagree with the adjudication if you think we drew the R1 line in the wrong place. That is the point of publishing it.

How new the role actually is

The widely repeated claim is that search interest in "finance engineer" grew 700%, with no query, geography or window attached. Posting dates are a harder measure and they are published in the ATS APIs, so we used those instead.

The median requisition in this corpus is 38 days old. None predates 2026. Measured on the first_published field for Greenhouse postings and publishedAt for Ashby, against a freeze date of 26 October 2026. The oldest posting went live on 27 January 2026 (Stripe, Bengaluru), and it is one of the five carried over from July. In July the median was 36 days and one posting dated from 2025.

41% of the requisitions were first published in the previous 30 days. 74% within 90 days. 89% within 180 days. 100% within a year. July read 46%, 76%, 95% and 100%. The shape is the same: essentially nothing older than a year. It carries the same caveat as before: job boards show open roles, so any corpus of live postings skews recent by construction. What is informative is the shape, not the raw recency.

How common the role is

2.0% of open finance-department roles are engineering roles. That is 1 in 51. Across the 418 boards scanned, 1,018 open postings sat in a finance-side department. 20 of them were engineering roles meeting the inclusion rule. In July: 26 of 987, or 2.6%. This is the denominator that no other published figure on this role provides.

27 requisitions across 22 employers, out of 418 employers scanned. Roughly 5% of the employers in the frame had at least one such role open (6% in July). The role is real and it is concentrated: Anthropic accounts for 4 of the 27, Block and Deliveroo 2 each. In July the largest employer was OpenAI with 5; it has one in October.

Where in the world the role exists

Europe still has the role at a rate close to the US. Of the 1,018 finance-department postings in the frame, 615 are US-based, 227 European, 48 Indian and 128 unspecified or elsewhere. Engineering roles as a share of those: 2.3% in the US (14 of 615), 1.8% in Europe (4 of 227). July read 2.6% and 1.9%. Location strings were classified by a slightly different keyword list this run, so the regional denominators are not strictly comparable with July's; the ratios are.

Four of the 27 requisitions are European: Helsing (Finance Data Engineer, Munich, carried over from July), Deliveroo (Engineer, Accounting Data and Infrastructure Systems, and Product Manager, Finance Engineering, both London) and Farfetch (Software Engineer, Microsoft Dynamics 365 Finance & Operations, Porto). In July no European posting used the words "finance engineering". Deliveroo now does: its Finance Engineering organization is shared with DoorDash and Wolt, which acquired it, and the London product manager role is described as part of "a world-class Finance Engineering (FinEng) organization across all global markets".

The Indian concentration did not hold. In July, 14.3% of Indian finance-department openings were engineering roles, driven by Databricks, Twilio and Stripe. In October it is 1 of 48 (2.1%): the Databricks and Twilio roles closed and were not replaced, and only Stripe's Bengaluru requisition remains. July flagged the figure as small-denominator and partly an artefact; that warning was warranted.

The vendor-side alternative

There are two ways a company can get this capability without a product that provides it: hire an engineer into its own finance function, or have a vendor deploy one. The second route has a name and a boom behind it, the forward deployed engineer, so we measured it on the same frame.

377 forward deployed engineer postings across 81 employers. Eight requisitions are finance-facing, against one in July. Counting titles containing "forward deployed" or "FDE" across all 418 boards. The July figure (348 postings, 66 employers) was measured on the 330 boards of the pre-European frame, so the totals are not comparable; the finance-facing subset is. It now includes Databricks (two financial services roles), OpenAI (Financial Services, New York), Faculty, and three roles at Harvey titled "Forward Deployed Finance Engineer", for Private Credit, Buyout and Real Estate, in its Go to Market organization. They are deployed into private equity and credit clients, and excluded from the corpus under R3.

One forward deployed engineer now sits inside a finance department. Coinbase's role is "embedded as a founding engineer within Finance", inside its Finance & Accounting department and the Office of the CTO, to "identify high-value workflow opportunities, rapidly prototype solutions using Coinbase's existing AI platform, and drive measurable adoption of AI tooling across the FP&A & Treasury org". It serves the employer's own finance function, so it is in the corpus. The title has moved from the vendor to the customer.

Widen the definition to every deployment-shaped role and finance accounts for 21 requisitions out of 787. Including solutions architects, solutions engineers, deployment strategists and implementation engineers gives 1,139 postings across 166 employers. Most finance-facing ones sit in field engineering, professional services or sales organizations attached to a platform sale. July's conclusion holds in direction, with a nuance: the deployment-engineer boom is starting to reach finance clients, through a few AI vendors, but it is still a rounding error next to the in-house route.

What the role is called

27 requisitions, 27 distinct titles. No two employers word it the same way. Exactly one requisition is titled plainly "Finance Engineer" (AssemblyAI), as in July (VulnCheck, now closed). "Finance Engineering" is increasingly the name of a team rather than a title: Anthropic's Engineering Manager and Staff Software Engineer roles, Klaviyo's Order to Cash Track Lead, Deliveroo's Product Manager and Block's Engineering Manager for Finance & Supply Chain Engineering. The others include Finance & Strategy AI Engineer, FP&A AI Architect, AI Acceleration Engineer, Financial Systems Engineer, Senior Finance Systems Architect, and Forward Deployed Engineer, Finance.

6 of the 27 titles contain no finance word at all, against 16 of 37 in July. Brex's "Data Engineer", Decagon's "AI Acceleration Engineer", IMC's "Software Engineer, BP&A", Render's and Temporal's analytics engineers and Farfetch's Dynamics 365 engineer all sit in a department carrying a finance label. Any count built on title matching still misses them, but fewer than in July.

Where the role actually sits

20 of the 27 requisitions sit in a finance-side department. 7 do not. Inside: Finance (Anthropic, AssemblyAI, Brex, Cognite, Decagon, Faire, Helsing, IMC, OpenAI), Finance & Accounting (Coinbase), Finance and Strategy (Samsara), Finance & Operations (Render), Finance / Data Engineering and Analytics (Temporal), Financial Analytics and Product (Financial Systems) (Deliveroo), 7420 Finance Systems (Stripe), and Technology / Orders, Logistics & Finance (Farfetch).

About a quarter of the roles are still owned outside finance, as in July. The 7 exceptions: Block's two roles in Engineering, Business Platform; Gusto's FP&A AI Architect in Security (its Enterprise AI team); Harvey in Product; Klaviyo in IT & Security; Lambda in Business Technology; and Lyft's finance technology internship, filed under Early Talent. That is 26% of the corpus, against 30% in July. Block, Harvey and Klaviyo placed the role outside finance in both runs.

Where a reporting line is spelled out, it is now as often outside finance as inside. Observed lines: "VP Finance" (AssemblyAI), "Vice President of FP&A" (Cognite), "Director, G&A Systems" (Klaviyo), "Director of Data Engineering and Analytics" (Temporal) and "Development Manager" (Farfetch). In July every stated line named a finance leader. Most postings still state no reporting line at all, which is why the department field carries the finding.

How senior the role is

Where a minimum is stated it runs from 3 to 12 years, with a median of 5. Stated in 20 of the 27 postings. Full distribution: 3, 3, 3, 3, 4, 4, 4, 5, 5, 5, 5, 5, 6, 8, 8, 8, 8, 10, 10, 12. July ran from 1 to 12 with the same median. The titles agree that the role is senior (Staff, Director, Engineering Manager, Architect), with one entry-level exception: Lyft's Business Systems Engineer internship in Finance Technology, the first internship in the corpus.

No posting requires evidence of past work. The claim that "finance just became a portfolio job" has circulated since August 2026. We added a field for it this run: does the posting ask for a portfolio, examples of automations put into production, or links to past projects? None of the 27 does, and a retroactive check on the 37 frozen July snapshots also finds none. The one appearance of "portfolio" in each run refers to a portfolio of systems or analytics products the hire will own, not to work the candidate must show.

What tools the postings name

Frequency across the 27 coded postings:

A blank July share means the tool did not reach the July table. Three of the seven Claude mentions are in Anthropic's own postings, where the model is both employer product and, in two of them, a user of the system being built; the other four name Claude Code as a development tool.

19 of the 27 postings (70%) reference AI agents, agentic workflows or LLMs in their requirements. In July: 21 of 37 (57%). Two postings describe finance systems whose users include agents, not only people. Anthropic's Finance Engineering roles: "Its users include Claude agents, which read and write through the same APIs and controls as people." Decagon's AI Acceleration Engineer will "build the interfaces that let others work with our systems through AI, including MCP servers, skills, and agent-facing tooling", and "AI-pill the rest of the company, giving them the MCP servers, skills, agents, and artifacts to automate their own work."

Model Context Protocol appears in 5 postings across 4 employers. Anthropic (both Finance Engineering roles: "Experience building with LLMs and agents, including MCP"), Decagon (above), Block ("Anthropic's Model Context Protocol (MCP)" among the technologies the team uses) and Gusto's FP&A AI Architect ("Familiarity with MCP-style integrations and AI governance frameworks"). In July: VulnCheck, OpenAI and ElevenLabs, none of which has a qualifying role open today.

AI coding tools are becoming a stated requirement. Four postings name Claude Code, Codex or Cursor as skills, against two in July. IMC asks for "hands-on experience with agentic coding tools (e.g. Claude Code, Codex, Cursor)"; Decagon for "hands-on experience with AI-native coding tools such as Claude Code, Codex, Cursor" used "to ship production solutions".

The systems underneath are still entirely conventional. NetSuite, Salesforce, Oracle and Workday dominate the platforms, Python and SQL the languages, and dbt and Snowflake the data layer. The novelty in these postings remains the agent and automation layer, not the stack.

What the postings pay

19 requisitions in the frame disclose a US base range. They span $130,000 to $405,000, with a median band midpoint of $231,500. Band floors run from $130,000 (IMC, Software Engineer, BP&A) to $320,000 (Anthropic). Band ceilings run from $180,000 to $405,000. The median midpoint is identical to July's to the dollar, on a set of requisitions that almost entirely turned over.

The full set of disclosed ranges. Where a posting gives several geographic zones, the top zone is shown and noted.

Gusto's band is $185,000–$205,000 in Denver and most remote locations.

The OpenAI comparison could not be repeated

In July, OpenAI had five engineering roles inside its Finance department, enough for a like-for-like comparison at one employer: they were paid 35% more than the rest of that department and within 0.5% of a product engineer. In October only one OpenAI requisition qualifies (Financial Systems Engineer, $240,000–$266,000). Two other engineering roles in OpenAI's Finance department are excluded under R4 (supply chain integrations and government support systems), so the comparison would not be like for like. We do not report it this quarter. The July result stands as a single-employer, single-quarter observation, unconfirmed.

Limitations

• The frame is a convenience sample of venture-backed technology employers. Greenhouse and Ashby are the applicant tracking systems of venture-backed tech, so scanning them finds venture-backed tech. Banks, insurers, industrials, healthcare systems and most mid-market companies run Workday, SuccessFactors or Taleo and are absent by construction. The pay figures reflect this most severely: the median band midpoint of $231,500 is a US tech number, not a market rate.
• The quarter-on-quarter change is within noise. 37 to 27 is the first measured delta at a frozen frame, and at these counts it is not statistically meaningful on its own. The series becomes informative with more runs, not with this one.
• The frame ages. It was frozen in July and is deliberately not updated, so employers founded or newly hiring since then are invisible, and boards that close are not replaced (421 live boards in July, 418 now). Any expansion will be reported separately and merged only at an annual rebase, with the break documented.
• Geographic coverage is two-sided but uneven. Asia beyond Indian delivery centres, Latin America and the Middle East are not sampled, and the Indian figures come from US employers' offshore sites rather than Indian employers.
• The 2.0% share applies to this frame only. It is the share of finance-department openings at scanned employers, not an economy-wide statistic.
• Counts are a floor. 27 is what we found and verified across 418 boards, not a projection of how many exist.
• Departments are employer-declared and not standardized. "Finance Systems" at Stripe, "IT & Security" at Klaviyo, "Early Talent" at Lyft and "Technology / Orders, Logistics & Finance" at Farfetch may describe similar functions under different names, and the last one is counted as finance-side because its label says so. We report the field as stated rather than normalizing it, and readers can re-adjudicate from the appendix.
• Adjudication involves judgment. The rules are written down and every exclusion count is published, but the boundary between "builds financial products" and "builds the finance function's systems" is genuinely blurry at fintechs. Stripe, Ramp, Qonto, BitGo, Mercury, Adyen, Coinbase, Gusto and Commure all produced matches we excluded on that basis this quarter. A reader who disagrees can add them back: it would roughly double the corpus and change its character entirely.
• Job postings expire, and faster than we estimated. The July run predicted roughly a quarter of the corpus would still be open after 90 days; 14% was. The page is refreshed quarterly against the frozen frame, and the counts should be read as snapshots of a moving population. Full-text snapshots of both runs are kept, so the underlying evidence survives the links.

Figures we excluded, and why

• "Search interest up 700%." Google Trends is reproducible only against a stated query, geography and window. The circulating claim specifies none of the three. We measured recency from posting dates instead, and now a quarter-on-quarter count at a fixed frame, which went down, not up.
• "Base salaries of $130K–$250K." No traceable source. Our 19 verified US disclosures reframe rather than confirm it: the floor happens to match this quarter (IMC, $130,000), but the ceiling is materially too low (ten postings top out between $256,500 and $405,000).
• "Finance just became a portfolio job." Not supported by posting text. None of the 64 postings coded across both runs asks for a portfolio or examples of past work. Interviews may; postings do not.
• "Hundreds of openings on LinkedIn." LinkedIn's result count expands through fuzzy title matching and counts each location separately. After collapsing multi-location duplicates and excluding product roles, the qualifying population across 418 boards is 27.
• "Figma is hiring finance engineers." Figma's board was checked in August outside the cycle and again in this run. Neither found an engineering-titled role in a Figma finance department, which still sits in Business Operations. In October that department does carry a "Strategic Finance, AI Innovation" role: AI work in finance at Figma, but not under an engineering title.
• ZipRecruiter, Glassdoor, Salary.com and Levels.fyi averages. Algorithmic aggregates over inferred title matches rather than readings of posting text.
• Fintech and financial-services engineering roles, excluded under R1 though a title search returns them in bulk. This quarter: Stripe's Accounting Products, Revenue and Financial Automation and Treasury for Platforms teams, Ramp's and Qonto's accounting engineers, BitGo's ledger systems, Mercury's treasury engineers, Adyen's financial data product, Coinbase's FinHub and institutional financing, Gusto's payments ledger, Commure's revenue-cycle ledger, and OpenAI's Applied AI "Codex for Finance" roles. These build financial products for customers. Different job, though some of them, such as Stripe's Accounting Products team, now describe their customers as "finance teams, and the agents that support them".

Corpus appendix

All 27 requisitions, with the department as declared by the employer in its ATS. Frozen 26 October 2026. "ns" means the posting states no figure. The 37 requisitions of the July run, with their archive links, remain in the dataset and the method folder.

Inside a finance-side department (20)

Outside a finance department (7)

Documented outside the frame (July run). Two postings cited in the July version of this page were found through web search rather than the board scan, and were excluded from every count so the ratios stay clean: Spring Health, Senior Finance Engineer (New York, $159,100–$179,000) and Greentube, Finance AI & Process Automation Engineer (Vienna, €60,000 collective-agreement floor; the posting returns HTTP 410 Gone and rests on its frozen snapshot).

Sources

Every figure on this page is read from one of the postings linked in the appendix, retrieved through:

• Greenhouse public job board API
• Ashby public job posting API

Non-ATS sources:

• Spring Health, Senior Finance Engineer (New York)
• Greentube GmbH, Finance AI & Process Automation Engineer (Vienna), no longer live
• CFO Connect, "What Is a Finance Engineer? The Emerging Finance Role for 2026", Luc Hancock, 11 June 2026. Cited for its characterization of the role as a skill layer added to existing positions, and for the CFO Connect Summit 2025 poll finding that 56% of finance professionals rank workflow automation as the most important AI-related skill for the next two years. Sample size not disclosed by the publisher.

Changelog

• 2026-10-26 (quarterly run, first time series point): Second run against the frame frozen on 27 July, unchanged at 1,061 identifiers. 418 live boards (421), 28,087 postings (27,278), 1,018 in a finance-side department (987). Requisitions 37 to 27, employers 26 to 22, share of finance openings 2.6% to 2.0%. This is the first comparable delta and it replaces the "+700%" claim as the page's measure of growth: on a fixed set of employers, the role did not grow this quarter, though the fall is within noise at these counts. Only 5 of the 37 July requisitions are still open (two of them retitled). Conclusions that changed, stated rather than corrected silently: (1) the Indian concentration (14.3%) did not hold and is now 2.1%; (2) stated reporting lines no longer all name a finance leader; (3) "no European posting uses the words finance engineering" is no longer true (Deliveroo); (4) the forward deployed route now has eight finance-facing requisitions, not one, including one inside a finance department (Coinbase); (5) the OpenAI pay comparison could not be repeated and is withdrawn from the current findings. Unchanged: median pay midpoint ($231,500), median seniority (5 years), roughly a quarter to a third of roles owned outside finance. Method changes: new exclusion rule R5 (lexical false positives) added without widening R1 to R4; new coded fields for evidence of past work (0 of 27, and 0 of 37 on the July snapshots) and AI coding tools; recency computed against the actual freeze date, which the script now reads at run time instead of from a hard-coded date; carry-over measured by ATS job ID rather than title, since two requisitions were renamed. The forward deployed counts were remeasured on the full 418-board frame (July's were on 330 boards). Next run: late January 2027.

• 2026-09-21 (link check): Every external link on this page was re-tested. One is genuinely dead: Greentube's Vienna posting now returns HTTP 410 Gone, and no Wayback capture exists because their careers site refuses the archiver. The citation is relabelled so the link is not presented as live; the quoted salary floor stands on the frozen snapshot and content hash recorded at retrieval. Expected attrition for a job-postings corpus, and the reason the archival pass exists. No finding on the page changes. Next scheduled refresh unchanged: late October 2026.

• Refresh cadence set to quarterly (next run late October 2026), against a frozen 1,061-token sampling frame so that counts stay comparable run to run. Ninety days is the longest interval that still leaves meaningful overlap between consecutive corpora: 76% of the current corpus was published within that window, so at six months two consecutive runs would share almost nothing and the delta would be uninterpretable. The queries, the adjudication rules and the exclusion counts are published in the Method section above, so the run is reproducible from this page alone.

• 2026-07-27 (European pass): Added 268 European candidate boards to the frame after the corpus was challenged on geographic coverage. 91 returned live boards, adding roughly 3,950 postings and 153 finance-department postings, but only two qualifying roles: Helsing (Munich) and Zilch (UK). Corpus 35 to 37 requisitions, 24 to 26 employers. The share of finance-department openings moves from 2.9% to 2.6%. New section on geography: Europe runs at 1.9% against 2.6% in the US, and no European posting uses the title "finance engineer". Also surfaced that 14.3% of Indian finance-department openings in the frame are engineering roles, driven by Databricks, Twilio and Stripe placing this function in Bengaluru.

• 2026-07-27 (recoding pass, at publication): The European pass added two requisitions to the corpus without re-running the full-text coding, so the tool, seniority and agent-reference sections were still reported on the 35 pre-pass postings while the corpus counts said 37. Helsing and Zilch were coded from their frozen snapshots and folded in: reporting-line split 24/11 to 26/11 of 37, experience minimum stated in 28 of 37 rather than 27 of 35 (range and median unchanged at 1 to 12 and 5), agent references 20 of 35 to 21 of 37 (57% either way), and the tool table recomputed on 37 (Python and SQL 54%, dbt 35%, Airflow 14%). Recency was also restated against the actual 27 July freeze date rather than the 25 July one it had been computed on: median age 34 to 36 days, and 49/76/97/100% within 30/90/180/365 days to 46/76/95/100%. Three stale counts from the pre-pass version were corrected in the Limitations and Excluded-figures sections (2.9% to 2.6%, 35 across 330 boards to 37 across 421). No finding on the page changes direction.

• 2026-07-26 (FDE contrast): Added a measured comparison against the forward deployed engineer route, on the same 330 boards: 348 FDE postings across 66 employers, of which exactly one requisition is finance-facing. FDE roles remain excluded from every other count, since they are employed by a vendor and deployed into a customer rather than serving the employer's own finance function.

• 2026-07-25 (archival pass): All 35 postings captured in full text with retrieval timestamps and content hashes; 32 of 35 independently archived to the Wayback Machine and linked from the appendix. Klaviyo's three roles could not be archived externally (its careers site and Greenhouse mirror both refuse the archiver) and are marked accordingly.

• 2026-07-25 (second revision): Rebuilt on a sampling frame. Rather than a hand-picked employer list, we tested 793 candidate board identifiers, scanned all 23,332 open postings on the 330 live boards, and filtered on employer-declared department. This produced a denominator (834 finance-department openings, of which 2.9% are engineering roles) that the previous versions lacked, and it surfaced 12 employers the earlier passes missed entirely. Added: posting-date recency as a measured replacement for the unverifiable "700% search growth" claim; a tool frequency table across all 35 coded postings; the OpenAI three-way compensation comparison; written adjudication rules R1 to R4 with published exclusion counts; and an expanded Limitations section. Corrected the pay set after finding that a decimal-band parsing error had distorted an earlier compensation median.

• 2026-07-25 (first revision): Corpus rebuilt on employer ATS APIs rather than title search, expanding it from 12 postings to 31. This changed a headline finding: the initial version reported that every posting stating a reporting line sat in Finance, based on five readable postings. A wider scan showed a meaningful minority outside it. Disclosed pay went from 2 postings to 17.

• 2026-07-24 (initial version): First corpus. 12 postings identified, 5 coded. Six widely circulated figures excluded and documented. No salary range published: only two postings disclosed a verifiable figure. Review initially set at six months rather than the usual twelve, because a job-posting corpus decays faster than survey data. Superseded: see the 27 July entry.

Further reading

• What these postings imply about the tooling underneath them: The Finance Engineer Job Description Is a Spec for Infrastructure Nobody Has Built
• Why an AI agent working on financial data needs a persistent model layer: Why Finance Agents Need a Model Layer
• The open convention for documenting a model's structure and conventions: The FINANCE.md Standard
• What AI adoption in the finance function looks like more broadly: AI in Finance: Statistics
• Where the time this role is meant to reclaim currently goes: Where FP&A Time Actually Goes

Compiled by Layerz.