AI Jobs Weekly (Nov 2, 2026): The Hiring Is Real in Exactly Three Places
What actually moved this week
Run date: November 2, 2026; every figure below carries its source and vintage.
Four numbers frame the week. First, Rung’s November 2026 census counted 391 open Forward Deployed Engineer (FDE) roles across 13 companies on public job boards (Rung, updated November 2, 2026). Second, Skillenai indexed 71 job postings titled “Applied AI Architect” over the 90 days ending 2026-10-29 (Skillenai). Third, the AI Evals Hiring Index counted 1,147 open roles across 171 companies as of 29 October 2026 (AutomationRoles.ai, 30 Oct 2026). Fourth, layoffs.fyi’s 2026 block reads 136,213 tech employees laid off across 337 tech companies year to date (layoffs.fyi, read 2026-11-02).
The honest frame is concentrated, not exploding. Three traps sit inside these numbers, and this post is built around them: a role count you will see two numbers for, a skill masquerading as a job title (2 titled postings in a re-run of 3,610, per Square1 Ai), and two layoff trackers that will never agree.
How This Guide Was Built
This guide was compiled on November 2, 2026. The two layoff trackers and the Rung FDE census were read the same day; the AI Evals Hiring Index is dated 29 October (published 30 October); Skillenai’s window ends 2026-10-29; the Square1 Ai study runs through 29 September 2026.
Editorial policy: figures are quoted as published. Where sources conflict, both are shown with their scope. Trackers with different coverage are never summed or averaged, and nothing is reconciled on the reader’s behalf. If a number’s basis is unclear, we say so.
Limitations, stated plainly: no source in this dossier describes an interview process or hiring loop for any of the three roles — only skills asked for in ads. No source provides Applied AI Architect pay. Job postings cited by third-party pages (specific Anthropic or OpenAI listings, for example) are examples those pages report, not independently verified live vacancies.
The layoff backdrop: two counters, two different truths
Two public trackers cover 2026 tech layoffs, and they do not match. Layoffs.fyi, read November 2, 2026, reports 136,213 tech employees laid off across 337 tech companies year to date. GoLayoffs, updated November 2, 2026, records 121 tech layoffs at 93 companies affecting at least 69,494 workers — about 227 a day (GoLayoffs).
Do not add these, average them, or pick the bigger one as “the real number.” The gap is a measurement difference, not a contradiction in the world: different coverage (337 companies versus 93), different event counts, different inclusion rules, and one tracker reads “at least.”
For context, layoffs.fyi’s prior-year blocks read: 2025 — 122,606 employees across 278 companies; 2024 — 152,922 across 551; 2023 — 265,660 across 1,194.
The cleanest year-over-year signal available here is GoLayoffs’ WARN-notices-only subset: large tech employers reported 28,745 layoffs in state WARN notices so far this year, 66% more than the 17,300 at the same point in 2025, with the same list of companies held constant in both years. Treat that as the most comparable YoY measure here.
The most recent dated layoff items, October 21–28, come from GoLayoffs, and no other source here adds weekly items: Dropbox (Oct 28, −190), Frontapp (Oct 27, −53), Microsoft (Oct 26), Instacart (Oct 25), Intel (Oct 22, −67), and Electronic Arts (Oct 21, −400).
RTO corner: buyouts plus five days in office
The week’s one return-to-office item is Skydance: on October 26 it offered voluntary buyouts across the newly combined company alongside a mandatory five-day return-to-office policy, announced by David Ellison and Ynon Kreiz; the mandate brings Warner Bros. Discovery in line with existing Paramount policy (Barrett Media). Caveat: Skydance is a media/entertainment employer, not a software company — this illustrates the buyout-plus-RTO pattern; it is not a tech-sector statistic.
Keep the FDE-specific on-site signal separate from any general claim: Rung’s census shows 81% of FDE postings mention travel or on-site work, and roughly 19% are flagged remote-friendly (Rung). Those are FDE figures, not a tech-wide RTO rate; no source here gives a current tech-wide rate.
Forward Deployed Engineer: 391 roles, and a second number you’ll see
A Forward Deployed Engineer embeds with a customer to build and deploy AI systems — a role Palantir invented around 2010, per AITraining2U’s explainer.
Demand, per Rung’s November 2026 census of public job boards: 391 open roles across 13 companies, moving from 392 to 391 in a month — up 0%. Title families inside the 391: Forward Deployed Engineer 236, Deployment Strategist 69, Forward Deployed Software Engineer 54, AI/Deployment Engineer 32.
Who is hiring, all 13: Databricks (115), Palantir (106), Salesforce (92) — together 313 of the 391 — then OpenAI (35), Scale AI (15), Cohere (13), Writer (3), Baseten (3), Sardine (3), Modal (2), Ramp (2), GitLab (1), Datadog (1).
Pay, from the 26 roles disclosing US bands: median total compensation ran $185K to $300K, topping out around $445K plus equity, with a floor around $132K.
Entry path: 51% of the open roles are mid-level or individual-contributor, not staff or principal. Skills, computed by Rung over 299 of the 391 postings with full descriptions: customer/client facing 99%, travel or on-site 81%, Python 79%, APIs and integrations 65%, cloud (AWS/GCP/Azure) 61%, JavaScript/React/TypeScript 59%, data/SQL/pipelines 53%, LLMs/GenAI 31%, US security clearance 29%, Kubernetes/Docker/infrastructure-as-code 9%. Geography: around 42% of roles are in the United States (165 roles), roughly 19% are flagged remote-friendly; the country counts after the US are UK 42, South Korea 11, Japan 11, France 11, Canada 11, India 9, Singapore 8, Spain 6, Australia 6, Germany 5.
One cross-link matters for later: in Square1 Ai’s 2026 study, 9 of 29 FDE ads (31%) asked for evals or evaluation (Square1 Ai).
The 224 vs 391 problem
You will see a second FDE count: a market scan in mid-2026 counted 224 open FDE positions across 39 AI companies — a jobsbyculture.com figure cited by AITraining2U, published 2026-06-10. The scopes differ: June versus November, 39 companies versus 13, different board coverage. The two counts are not mergeable, and these sources do not explain the discrepancy. For this week, trust the November 2 census — it is the fresher read on a defined company set.
Pay has the same shape of trap. AITraining2U reports FDE total compensation “from $300,000 to $1.2 million a year at the top of the market,” with no stated sample or date basis beyond mid-2026. Rung’s top end is around $445K plus equity across 26 disclosed US bands in November 2026. Both are reported as given; neither is reconciled here.
One more flag: AITraining2U’s mid-2026 claim that FDE is “the most-hired-for engineering role at OpenAI, Anthropic and Google” is not corroborated — neither Anthropic nor Google appears in Rung’s November list of 13 companies. Attributed, not adopted.
Applied AI Architect: 71 postings, zero public pay data
The Applied AI Architect is “the technical advisor who helps an enterprise evaluate a frontier model and designs how it gets deployed, working from technical discovery through to a scalable architecture” (CTAIO). It is a pre-sales and advisory seat: the Architect designs the deployment; the FDE builds it afterward. The distinguishing skill is knowing where a proof of concept stops scaling.
Demand is the thinnest of the three roles: Skillenai indexed 71 postings titled “Applied AI Architect” over the 90 days ending 2026-10-29, concentrated in New York City, London, Munich, Mumbai, and Washington.
Pay: no source in this dossier reports Applied AI Architect pay bands. That is a fact about the evidence, not a gap to fill with an estimate.
Companies: Anthropic and OpenAI both hire the title. At Anthropic the role sits inside the Applied AI organisation and works across Sales, Product and Engineering, with variants including Public Sector, Enterprise Tech, Industries and APAC postings, plus a Head of Applied AI Architects and a Beneficial Deployments leadership seat. OpenAI’s San Francisco posting frames the role as “the CTO of your book of business” — a senior technical owner for a named portfolio of enterprise customers. Caveat: these are postings the cited page reports, not independently verified live vacancies.
Skills most often mentioned: Python, data science, machine learning, LLM frameworks, evaluation frameworks (Skillenai). Note the crossover — “evaluation frameworks” connects this seat to the evals cluster below. Entry path: the sources frame this as senior and advisory; no entry-level signal exists.
Evals: the skill that barely exists as a title
Here is the tension, stated hard. In Square1 Ai’s re-run of 3,610 postings on 29 September, only 2 had “evaluation” in the title — yet in its count of 315 AI job ads from July to September 2026, 73 (23%) asked for evals or evaluation (Square1 Ai). The study’s own advice: “Searching for ‘evals engineer’ will find almost nothing.” The titled tracker agrees: Glyph records “Evals Engineer” as first seen Dec 2025, with 14 open roles, 0% of HN hiring posts, and a remote option in 14% of postings (Glyph).
Where the skill is actually asked for (Square1 Ai, from its 19 September headline sample and 29 September re-runs): forward deployed engineer ads, 9 of 29 (31%); sales or solutions engineer, 10 of 49 (20%); software engineering ads that name AI coding tools, 11 of 67 (16%); product manager, 12 of 109 (11%). In the 315-ad set, Python appeared in 38% and agents in 39%.
Broad demand, measured three different ways that must not be merged. The AI Evals Hiring Index counted 1,147 open roles across 171 companies as of 29 October 2026; its biggest hirers are Anthropic (99), OpenAI (55), Amazon (39), Waymo (39), Scale AI (36), and xAI (32) (AutomationRoles.ai). DataPayIndex, a third source cited via AutomationRoles.ai, found 1,248 data and AI postings at 683 companies mentioning LLM evaluation — about 9.5% of all the data and AI postings it tracks. Mettlespark, also cited via AutomationRoles.ai, read 46,198 tech job ads across the US and Canada from May to October 2026: LLM evaluation appears in 28% of AI Engineer ads and 23% of Machine Learning Engineer ads, and when an ad names it, 79% list it as required rather than nice to have.
Pay — three medians, three scopes, never averaged. The Index: 553 of its 1,147 roles publish a salary band, with a median band of US$220k to US$330k in base pay; equity is not included. DataPayIndex: US median US$216k, senior roles at about US$204k, staff or lead roles around US$239k. Glyph: middle half $212k to $238k, median $232k base, from 10 postings that state a range. Caveat from the source: the headline bands are mostly US frontier lab and big tech numbers, and pay outside the US is lower.
Named roles in the dossier include ServiceNow (Staff and Senior ML Engineer, Agent Eval Platform), Apple (ML Engineer, Safety Human Evaluation, Cambridge MA), and Waymo’s 2027 summer internships in simulator realism evaluation.
Entry path — the five sourced skills: Python; enough statistics not to fool yourself; rubric writing and labelling; calibrated judges; and evals in the build (CI gates plus monitoring). You will be hired for this skill inside another title, not under it.
The three roles side by side
| Dimension | Forward Deployed Engineer | Applied AI Architect | Evals |
|---|---|---|---|
| Demand evidence | 391 roles at 13 companies (Rung, Nov 2, 2026); flat month (392→391); a separate mid-2026 scan counted 224 across 39 companies — different scope, not merged | 71 titled postings in the 90 days ending 2026-10-29 (Skillenai) — thinnest evidence | Widest reach, weakest title: 23% of 315 AI ads mention it (Square1 Ai); 1,147 roles at 171 companies (Index, Oct 29); only 2 titled postings in 3,610 and 14 open titled roles (Glyph) |
| Pay evidence | Median total compensation $185K–$300K across 26 disclosed US bands; top ~$445K plus equity; floor ~$132K (Rung). A separate claim of $300K–$1.2M at the top of market stands unreconciled | No pay data in these sources | Three separate medians: US$220k–US$330k base band (Index, 553 bands, equity excluded); US$216k (DataPayIndex); $232k median base (Glyph, 10 postings) |
| Entry path | Most accessible on paper: 51% mid-level or IC; Python 79%, APIs 65%, cloud 61%; travel/on-site 81%; US clearance required in 29% | Senior, advisory framing; “CTO of your book of business”; no entry-level signal in sources | Enters through other titles; 79% of ads naming it call it required; Waymo 2027 summer internships are an early signal |
| Remote signal | ~19% flagged remote-friendly; 81% mention travel or on-site | Not quantified in these sources | 14% remote option (Glyph) |
| Cross-links | 31% of FDE ads ask for evals (9 of 29, Square1 Ai); FDE builds what the Architect designs | Skills overlap with evals (“evaluation frameworks”); Architect precedes FDE at Anthropic’s Applied AI org | Highest mention rate sits inside FDE ads (31%) |
Your next 30 days: three moves
Move 1 — Build and publish one eval artifact: a grader, a rubric, and a CI gate. Basis: Square1 Ai’s five employer-requested skills, and Mettlespark’s finding that 79% of ads naming LLM evaluation list it as required. Because only 2 of 3,610 postings carry the title, you will be hired for this skill inside another one — the published artifact is the proof you attach to ML-engineer and FDE applications (Square1 Ai).
Move 2 — Apply to FDE roles at the three employers holding 313 of the 391 openings, and lead with customer-facing evidence. Basis: Rung’s census puts Databricks (115), Palantir (106), and Salesforce (92) at 313 of 391; 99% of descriptions emphasize customer-facing work, 79% ask for Python, 65% for APIs and integrations. With 51% of roles mid-level or individual-contributor, this is the most level-accessible of the three roles this month (Rung).
Move 3 — Repackage your systems experience as an Applied AI Architect narrative — “where a proof of concept stops scaling” — and target the Anthropic and OpenAI title. Basis: 71 titled postings in 90 days, concentrated in New York City, London, Munich, Mumbai, and Washington, with the “CTO of your book of business” framing. Hold the ambition and the caveat together: no pay data exists for this title in these sources (Skillenai, CTAIO).
FAQ
How many Forward Deployed Engineer jobs are open right now? 391 across 13 companies, per Rung’s census updated November 2, 2026 — flat on the month (392 to 391). You may also see 224 across 39 AI companies; that is a mid-2026 scan cited by AITraining2U with a different date, company universe, and board scope, and the two are not comparable.
What do FDEs earn? Across the 26 Rung roles disclosing US bands, median total compensation ran $185K to $300K, topping out around $445K plus equity, with a floor around $132K. AITraining2U separately claims $300,000 to $1.2 million “at the top of the market” without a stated sample; the conflict is reported, not resolved.
What does an Applied AI Architect earn? No source in this dossier provides pay bands for the title. Full stop.
Is “evals engineer” a real job title? Barely. Square1 Ai found 2 titled postings in 3,610; Glyph tracks 14 open “Evals Engineer” roles, first seen Dec 2025. Yet 23% of 315 AI ads asked for the skill. Do not search by the title — look inside FDE, sales or solutions engineer, software engineering, and product manager ads, where mention rates run 31%, 20%, 16%, and 11% respectively.
Are layoffs worse in 2026? The two trackers cannot be combined, so there is no single answer from them. The cleanest comparison is GoLayoffs’ WARN-only subset: 28,745 layoffs so far this year versus 17,300 at the same point in 2025 on the same company list — 66% more.
Can these roles be remote? FDE: roughly 19% of postings are flagged remote-friendly, and 81% mention travel or on-site work (Rung). Evals: Glyph shows a remote option in 14% of postings. Applied AI Architect: no source quantifies remote availability for that title.
Do I need a security clearance? For FDE work, 29% of postings require a US security clearance (Rung, computed over 299 of 391 postings). Nothing in the dossier addresses clearance for the other two roles.
The Bottom Line
AI hiring in the week of November 2, 2026 is real but narrow, and it is happening in exactly three places. Forward Deployed Engineer has the strongest demand evidence — 391 open roles at 13 companies — but 313 of those sit at Databricks, Palantir, and Salesforce, and the month was flat (392 to 391). Applied AI Architect is real but thin: 71 titled postings in 90 days, five cities, two named hirers, and zero public pay data. Evals is the widest-reaching of the three — 23% of sampled AI ads ask for it, and 79% of ads that name it call it required — yet it almost never appears as a title (2 of 3,610; 14 roles on Glyph), so you enter through FDE, solutions engineering, software engineering, or product roles.
Meanwhile the layoff backdrop is genuinely worse on the cleanest year-over-year measure available — the WARN-notice subset, up 66% on a constant company list — even though the two headline counters (136,213 and at least 69,494) can never be added together or averaged into one number.
If you are mid-level, customer-facing, and can write Python, the highest-probability 30-day play is an FDE application to Databricks, Palantir, or Salesforce with one published eval artifact — grader, rubric, CI gate — attached. If you are senior and advisory, the Architect title at Anthropic or OpenAI is worth targeting with eyes open about the missing pay data. And if you build evals into your existing engineering work, you are building a skill that 79% of the ads naming it call required.
← Back to all posts
