Case Study · Hiring Signals
Kerr Highland ran its hiring-signal read on three companies that anchor competitive-intelligence work in AI: Apple, OpenAI, and Tesla, with Meta and Google alongside as calibration. Every table below is a measured result from the firm's postings corpus.
Of OpenAI's 125 postings in the corpus, 95 land in 2025–26. Their language separates cleanly from everything earlier. A forecasting-science cluster appears from nothing: forecasting in 14 late postings, scientist, forecasting in 7, global forecasting in 5, with zero occurrences in the company's 2023–24 postings. Beside it, a user-operations and integrity function: user operations in 12 postings against zero from every other company in the corpus, with integrity measurement, AI fairness, and bias roles alongside. Teams form before products ship; these two announced themselves through recruiting.
| Term | OpenAI postings | Rest of corpus | First appears |
|---|---|---|---|
| user operations | 12 | 1 | 2025 |
| forecasting | 14 | 90 of 122,377 | 2025 (at OpenAI) |
| global forecasting | 5 | 0 | 2025 |
| integrity measurement | 4 | 0 | 2025 |
| ai fairness | 4 | 0 | 2025 |
Tesla's corpus footprint is smaller, 86 postings, and its sharpest signal sits where robotics-focused intelligence work looks: Optimus is the company's single most distinctive hiring term, appearing in its titles and nowhere else in the corpus. The 2025–26 window also shows the software tilt of the vehicle business: software engineer rose to 18 late postings with technician and vehicle engineering alongside.
Apple's distinctive language is a single loud story: the AI/ML organization around Siri. AIML appears in 64 postings, Siri in 16, machine learning innovation titles alongside, forming the most concentrated company-specific vocabulary in the study. The emergence table adds a quieter signal with direct relevance to advertising-market intelligence: ads appears in 13 Apple postings in 2025–26, a term absent from Apple's 2023–24 window.
Run the same read on Meta and the model reports ML software-hardware co-design, infrastructure experimentation, and technical-leadership hiring. A reader inside either company can grade that answer against the reality they know.
Calibration: the method scored against ground truth the client already holds
Meta's 627 postings read as ML infra, SW/HW co-design, infra experimentation, and technical leadership; Google's 755 read as Workspace, Google Ads, Cloud, quantum AI, with mobile and search engineering surging in 2025–26. These two runs exist so a reader can judge the instrument against companies whose reality they can check.
Corpus: Kerr Highland's licensed postings database; window January 2023 through July 2026; job-title language only. Company reads: differential term analysis (smoothed share ratios, company versus the rest of the corpus), minimum four postings per term. Emergence: terms present in a company's 2025–26 postings and absent from its 2023–24 postings, minimum three postings. Sample sizes: Apple 432, OpenAI 125, Tesla 86, Meta 627, Google 755. Small samples are reported as counts, and no claim rests on fewer than four postings. Full run file available.
Most strategic hires carry a generic title, an engineer of some flavor, and disclose themselves in the requirements: the unusual skill, the specific hardware, the domain experience nobody asks for by accident. The production monitor reads full descriptions, where a software engineer wanting low-latency sensor-fusion experience is a wearables team regardless of what the title says. That titles alone caught OpenAI's forecasting cluster is the measure of what the full-description read returns. As an evergreen monitor, it runs weekly on the companies a client names and surfaces formation as it happens. The precedent runs back to 2012, when the same logic read a cluster of nephrology hires at Roche as an oncology program targeting the kidney. The announcement came quarters later.