AI Hiring Brief · 2026-07-26

32 frontier AI labs · 37 roles added in 24h · agihunt.info/en/jobs

Today at a glance

Pace pulls back sharply: Daily additions across all tracked companies total roughly 37 new roles — well below the week's running average — suggesting a normal mid-cycle lull rather than any strategic shift.
Anthropic concentrates on RL infrastructure and frontier-safety research: Four of its eight new roles land in AI Research & Engineering, spanning Code RL, Environments Infrastructure, a "Takeoff Intel" research scientist, and an Agents research engineer — an unusually dense cluster for a single day.
NVIDIA builds a staffed Local AI team in Pune: Three roles opened simultaneously in India — two senior system software engineers plus an engineering manager, all under the "Local AI" banner — establishing a management-layer node for on-device and edge inference software, not a trial posting.
OpenAI tilts toward GTM and commercial infrastructure: Three Go To Market roles, two recruiting coordinators, an ads revenue accounting manager, and a first physical-security post in Singapore make up the bulk of today's additions — a commercialization support stack more than a research push.
Cursor opens its first government-facing role: "Strategic Partnerships & Growth Lead — Civilian Government" marks the AI coding tool's explicit entry into the U.S. public-sector market.

Company moves

OpenAI added 11 and closed 5. Go To Market drove the day: "Applied AI Engineer, GTM Growth Engineering," "Early Access Deployment Engineer," and "Data Scientist, GTM" collectively point toward scaling early-access products into paid adoption. Two People-side hires — "Recruiting Coordinator" and "Recruiting Coordinator Lead, Applied" — signal accelerating headcount growth specifically in the Applied org. On the finance side, "Sr Structured Finance Execution Lead" and "Senior Manager, Ads Revenue Accounting" suggest advertising revenue has grown large enough to warrant dedicated accounting management. Infrastructure gets one "Data Engineer, CPU & Storage" in the Scaling department. A "Travel Security Manager - APAC" in Singapore is the first physical security presence OpenAI has posted in Asia Pacific.

Anthropic added 8 and closed 1. Research engineering dominated: "Staff Software Engineer, Code RL" targets reinforcement learning for coding capabilities; "Staff Software Engineer, Environments Infrastructure" builds the training sandboxes that RL depends on; "Research Scientist, Takeoff Intel" is the most distinctive role of the day — in AI safety circles, "takeoff" refers to a period of rapid, potentially discontinuous capability gains, and Anthropic is now staffing dedicated intelligence work around it. "Research Engineer, Agents" extends the agents research line visible across last week's additions. On the business side, "Senior Manager, Technical Accounting - M&A and Investments" suggests active deal flow, while "Regional State and Local Affairs Lead, Southern States" extends Anthropic's policy footprint from the federal level into southern U.S. states.

NVIDIA added 7 and closed 1. The Local AI cluster in Pune is the structural story: "Senior System Software Engineer - LocalAI," "Senior System Software Engineer - Local AI," and "Manager, System Software Engineering - Local AI" opened together — a management layer over two senior engineers signals a team charter, not an experiment. "Senior Deep Learning Scientist, Multimodal Agentic RL" adds cross-modal agentic research capacity. "Machine Learning Engineer, AI Safety" continues a safety-hiring thread visible over the past week. "ASIC Design and STA Engineer" maintains steady chip-design pipeline hiring.

Microsoft AI added 6 and closed 17, continuing a pattern of high-volume churn at scale (532 open, 203 added over seven days). Today's new roles tilt toward cloud field deployment: "Cloud Solution Architect - Factory" covering Vancouver, Greater Toronto, and Ottawa targets industrial digitization; "Cloud Solution Architecture Manager-Azure" in Brasilia extends Latin American presence. Core engineering sees routine additions at Principal and Software Engineer II levels in Redmond.

Apple AIML added 2, both in Cupertino. "Senior Machine Learning Engineering Manager, Responsible AI and Safety - AIML/SWE" is a management-level responsible-AI hire — combined with data governance roles from last week, Apple is assembling a leadership layer for AI ethics and safety rather than individual contributors. "AIML Data Operations - Senior Manager, Communications, Engagement and Learning & Development" is an internal org-building role for the data operations team.

Perplexity added 1: "Member of Technical Staff (AI Software Engineer, Agents)" in San Francisco, continuing its agents-capability buildout.

Cursor added 1: "Strategic Partnerships & Growth Lead — Civilian Government," posted as Remote. This is the first explicit government-market role from an AI coding tool, and the Remote designation means it is unconstrained by geography — a deliberate, broad public-sector entry signal.

Lovable added 1: "Enterprise Account Executive, London" covering London and Stockholm — European enterprise sales coverage expanding.

ByteDance Doubao/Seed (字节跳动豆包/Seed) closed 31 roles today with zero additions, a notable single-day reduction from its 964-role open pool. Meituan LongCat (美团·龙猫) closed 10. Both figures likely reflect normal pipeline clearing — fulfilled positions or expired postings — rather than structural retreat, given their still-large open counts.

Google DeepMind, Meta AI, Cohere, Scale AI, Cerebras, SpaceXAI, and the remaining China-based labs tracked here posted no new openings today.

What it signals

The seven-day picture is dominated by Microsoft AI's 203 additions, a number that dwarfs every other tracked player and reflects the scale of field deployment: most of those roles are cloud solution architects and customer-facing engineers rather than researchers. OpenAI's 75 and NVIDIA's 45 over the same window sit closer to each other than to Microsoft's pace, but their compositions diverge sharply — OpenAI's additions increasingly split between commercial operations (GTM, finance, recruiting) and core AI engineering, while NVIDIA's spread across silicon (ASIC), system software (Local AI), and research (multimodal agentic RL) traces a hardware-through-stack strategy. Anthropic's 40 additions are concentrated almost entirely in AI Research & Engineering and public policy, making its per-hire technical density the highest of any lab in the window.

The Takeoff Intel hire at Anthropic is the most pointed signal of the day. Staffing a dedicated research scientist role around monitoring and analyzing rapid AI capability progression is qualitatively different from having safety researchers think about it alongside other work. It mirrors, in a different idiom, the Frontier Strategy and Governance roles that appeared at Google DeepMind last week: leading labs are converting what was previously theoretical concern into an operational function with named owners. Whether that reflects genuine internal belief that capability thresholds are approaching or is partly a posture for regulators, the organizational signal is the same — it is now considered a job, not a background assumption.

NVIDIA's Pune cluster deserves attention beyond raw headcount. Three roles with a management layer means a team charter with its own reporting line and presumably a roadmap. NVIDIA's on-device inference ambitions have been visible in product positioning for some time; translating them into a geographically distinct engineering org in India suggests the software scope is large enough to require its own structure, separate from the main GPU software stack in Santa Clara. Apple's "Responsible AI and Safety" engineering manager hire follows the same logic at the governance layer — the phase of individual contributors defining the function has passed; senior managers now own it.

On the commercial and geographic frontier, today's additions from Cursor (U.S. government), Lovable (European enterprise), OpenAI (APAC physical security), and Anthropic (southern U.S. state policy) sketch a consistent pattern: the leading AI products are past the phase where technical superiority alone sells. They are now building the government relations, enterprise sales, regional compliance, and physical-security functions that large institutional customers require before procurement can begin. China-based labs are notably absent from this international expansion hiring, with ByteDance Doubao/Seed and Meituan LongCat both running net-negative today on pure closure volume. Their open-role totals remain substantial, but the international sales and policy infrastructure visible among global leaders has not yet surfaced in their posted openings.