AI Hiring Brief · 2026-07-11

19 frontier AI labs · 92 roles added in 24h · agihunt.info/en/jobs

Today at a glance

Google DeepMind goes hands-on with robotics hardware: Five physical-hardware roles — "Mechanical Engineer, Robotics," "Electrical Engineer, Robotics," "Electrical Engineer, Firmware, DeepMind," "Engineering Technician, Robotics," and "Engineering Technician, Mechanical Assembly, Robotics" — all landed in Cambridge, MA in a single day, covering the full mechanical-electrical-firmware-assembly stack. This is an unusually concentrated hardware push for a lab best known for software research, signaling a shift toward owning its own robot platform rather than relying on third-party hardware.
Anthropic layers security depth, with an M&A signal embedded: Five of today's 9 new roles fall under security or Safeguards. "Staff+ Application Security Engineer - M&A" is the sharpest signal: a dedicated M&A security hire implies active deal diligence, not contingency planning. "Country Lead, Data Center Security" in Sydney marks Anthropic's first identifiable data center security leadership in Australia.
OpenAI builds an enterprise backbone in a single day: Four finance roles — "Treasury Manager," "Order Management & Billing," "Tech Lead, Finance & Supply Chain Engineering," and "Deal Desk Regional Lead, APAC" in Singapore — combined with a "Technical Program Manager, Hardware Chips Development" and "Lead Thermal Simulation Engineer," sketch an organization crossing from startup operations to the infrastructure required to run a major hardware and commercial program at scale.
Microsoft deploys a six-country AI partner sweep across Southeast Asia: A single "Partner Solution Architect: Data & AI" posting spans Malaysia, Singapore, Thailand, Vietnam, the Philippines, and Indonesia simultaneously — a rare single-JD regional blitz that signals enterprise AI sales acceleration into Southeast Asia at meaningful scale.
NVIDIA spans silicon and robotics in parallel: "Senior AI Verification Chip Design Engineer," "Senior ASIC Design Engineer - DFX" (design-for-testability), and "Senior Signal & Power Integrity Engineer" address the reliability layer of AI-specific silicon, while "Senior Manager, Software Engineering - Robotics Manipulation" opened across six locations — a buildout that directly mirrors what DeepMind is doing on the hardware side.

Company moves

Google DeepMind added 12 roles today. Five cluster in Cambridge, MA on physical robotics hardware, with no precedent in recent posting history for this kind of single-day concentration across mechanical, electrical, firmware, and assembly disciplines. The remaining seven span foundational research ("Staff Research Scientist in RL," "Research Scientist, Deep Learning"), tool-use and multimodal reasoning ("Research Engineer, Tool Use," "Research Engineer, Multimodal Reasoning For Information Literacy"), safety ("Senior Research Engineer, GenMedia Safety"), human understanding ("Research Engineer, Human Understanding, DeepMind"), and a consumer-product role ("Senior Software Engineer, Android, Gemini App").

Anthropic posted 9 new roles, with security and Safeguards accounting for five. "Engineering Manager, Safeguards Interventions" and "Staff+ Software Engineer, Safeguards ML Infrastructure" together indicate Safeguards is building its own ML engineering organization to support the policy and research functions. The "Standards Editor" in Communications is likely oriented toward AI governance text and safety commitments rather than conventional editorial work.

OpenAI added 16 roles across three threads. Finance and operations: four roles covering treasury, billing, supply chain, and an APAC deal desk seat in Singapore. Hardware: "Technical Program Manager, Hardware Chips Development" and "Lead Thermal Simulation Engineer" for heat management at data center scale. Go-to-market: "Account Director, Oil & Gas" deepens vertical industry coverage, "North America Co-Sell Leader" builds ecosystem channels, "Competitive Intelligence Lead" formalizes market awareness. The "Principal Software Engineer, Codex Cyber" in San Francisco is the clearest product signal: Codex is extending into cybersecurity as a named vertical.

Microsoft AI was the highest-volume mover with 29 additions, though it also closed 35 roles — net slightly negative on the day. Beyond the Southeast Asia partner architect blitz, standouts include "Senior Director, Datacenter End-to-End Build Delivery" across multiple US cities, "Member of Technical Staff - Applied AI Lead, Health" in London (a specialist health AI role suggesting vertical-product development outside the US), and continued security hiring via "Senior Security Researcher" and "Threat Intelligence Manager."

NVIDIA added 12 roles. The chip-layer cluster — "Senior AI Verification Chip Design Engineer," "Senior ASIC Design Engineer - DFX," "Senior Signal & Power Integrity Engineer" — reflects the maturation phase of AI chip development where verification discipline becomes the constraint. "Systems Software Engineer, Autonomous Systems Mapping - New College Graduate 2026" signals continued investment in the autonomous systems pipeline even at the entry level.

Meta AI added 4 roles, including "Software Engineer, AI Specialist - Wearables AI (Technical Leadership)" in Burlingame (Meta's AR hardware hub) and "Research Scientist, World Modelling | Chercheur scientifique, modélisation du monde" in Montreal — the latter a foundational research hire that fits Meta's long-running push toward predictive and embodied AI.

Perplexity added 3 roles: "Design Systems Lead" and two "Member of Technical Staff" positions targeting enterprise adoption and integrations. Both engineering roles are filed under AI Products, suggesting Perplexity is structuring a more deliberate enterprise motion rather than relying on organic consumer-to-business spillover.

Cursor added 2: "Director, Partnerships SI" (systems integrators) and "Engineering Manager, ML" — a pairing that simultaneously scales the ML core and builds the partner distribution layer around it.

Moonshot AI (月之暗面 Kimi) added one role: "GPU算力采购经理" (GPU computing procurement manager), Haidian District. Among all Chinese labs tracked today, this is the only addition — and it points to GPU supply chain investment rather than research or product headcount.

Closures without additions: Baidu AI (百度 AI) closed 20 roles with zero additions; Meituan LongCat (美团·龙猫) closed 7; ModelBest (面壁智能) and Lovable each closed 2. Mistral AI, DeepSeek (深度求索), Zhipu AI (智谱 AI), and StepFun (阶跃星辰) recorded no movement today.

What it signals

The most striking theme in today's data is the AI industry's pivot toward physical infrastructure. Google DeepMind's five-role Cambridge robotics hardware cluster is not a research experiment; the combination of mechanical engineering, two electrical engineering specialties, firmware, and two technician tracks is the organizational shape of a team building and assembling hardware at bench scale. NVIDIA simultaneously opening a "Senior Manager, Software Engineering - Robotics Manipulation" across six sites reinforces this: the frontier of competition is shifting from model benchmarks to the supply chains — chips, thermal management, robot embodiment, and hardware assembly — that determine who can actually deploy capable systems at scale.

The 7-day figures reveal a two-speed industry. Global labs are posting at institutional rates: OpenAI at 787 additions over seven days, Microsoft AI at 637, NVIDIA at 613, Anthropic at 433, Apple AIML at 369. Chinese labs in this dataset run at substantially lower absolute volumes, with Zhipu AI leading at 237 open roles and 237 added over seven days — but posting nothing today. Baidu AI's 20 closures with zero additions is the sharpest domestic signal: a large incumbent reducing headcount without replacement, a pattern that suggests strategic retrenchment rather than a hiring pause.

The simultaneous security buildups at Anthropic and Microsoft AI deserve separate attention. Anthropic's investments are inward-facing — Safeguards ML infrastructure, intervention engineering, and M&A security diligence all reflect an organization trying to expand safely under its own commitments. Microsoft's additions are externally oriented — Threat Intelligence Manager and Senior Security Researcher fit a cloud provider whose enterprise customers face escalating regulatory requirements. The two postures are different, but the underlying pressure is the same: at current AI adoption levels, security and governance are becoming binding constraints on further growth.

Productization geography is expanding. OpenAI seated an APAC deal desk role in Singapore. Anthropic hired its first Australian data center security lead. Microsoft deployed one job description across six Southeast Asian markets in a single day. Perplexity hired for enterprise adoption and integrations rather than search improvements. These are not exploratory pilots — they are the staffing signatures of companies converting capability leads into contracted revenue in markets outside their original Bay Area base, and the pace of that expansion appears to be accelerating.