AI Hiring Brief · 2026-07-24

32 frontier AI labs · 103 roles added in 24h · agihunt.info/en/jobs

Today at a glance

• Embodied AI moves from research to data collection: Google DeepMind posted a "Specialist Humanoid Tele-Operator" in Mountain View — a role whose primary job is to operate humanoid robots and generate real-world training data. On the same day, NVIDIA opened a "Senior Software Engineer - Applied Research in Robotics" in Bengaluru, and OpenAI added "Senior Counsel, Commercial (Robotics)" in San Francisco. Three major labs, three different layers of the robotics stack, all on the same day.
• Microsoft AI's week-long surge continues: 38 additions today against 43 closures, with a rolling 7-day total of 183 new roles — the highest weekly volume across all 32 tracked companies. Roles span the MAI Superintelligence Team in London and Zurich, vector search infrastructure, and enterprise sales from South Africa to São Paulo.
• OpenAI extends its AWS channel globally: "AWS Specialist Seller, Strategic Pursuits" appeared in Singapore, and "AWS Specialist Sellers, Strategic Pursuits" simultaneously in London and Munich — a single commercial motion deployed across three timezones, signaling that the AWS partnership is entering systematic field execution rather than pilot territory.
• ByteDance Doubao/Seed posts the day's largest closure count: 68 roles closed against only 3 added, a net of −65. All three new openings are safety and evaluation functions: "大模型风险治理运营-大模型安全群组" (large-model risk governance operations), "AI大模型评估专家（生产力方向）" (LLM evaluation expert, productivity track), and "AIGC系统评测专家 - TikTok研发" (AIGC system evaluation expert, TikTok R&D).
• Enterprise AI field deployment goes emerging-market: Meta added four "Forward Deployed Engineer, Enterprise AI" roles in a single day — two in São Paulo and two in Singapore. Paired with Anthropic's new "Head of International Customer Success" in London and OpenAI's Singapore AWS seller, the geographic spread of enterprise AI go-to-market is visibly widening.

Company moves

Microsoft AI led all companies today with 38 additions, continuing a week that has produced 183 new postings. The standout is "Member of Technical Staff, Data Research Engineer - MAI Superintelligence Team" opening in London and Zurich — the superintelligence team is now building a European research footprint in two high-talent cities. A "Principal Software Engineer - (Vector Search/Information Retrieval)" role in the US strengthens retrieval infrastructure, consistent with the RAG and Copilot product lines. An "Executive Director Enterprise and Commercial Sales - South Africa" in Johannesburg represents the first senior commercial appointment to the African continent this tracker has logged. The volume is high but so is turnover: 43 closures in the same 24 hours suggest active role rationalization alongside expansion.

OpenAI added 13 roles (66 over seven days) and closed 9. Beyond the robotics counsel and AWS sellers noted above, "Offensive Security Agent Engineer" opened across five locations including remote — a specialist agentic red-team role that OpenAI is building in-house rather than outsourcing. Two growth marketing roles ("Growth Product Marketing Manager" and "Growth - Lifecycle Lead") both sit under Applied AI and target San Francisco and New York, pointing to a push on consumer-facing growth loops tied to the API product line. "AI Success Engineer" in Sydney is the first Australian customer success posting in recent windows.

Anthropic added 9 and closed 14, a net of −5. The day's clearest signal is the double investment in GTM Systems: "Staff Software Engineer, GTM Systems" and "Technical Program Manager, GTM Systems" posted the same day suggest the go-to-market tooling stack is being productized internally. "Head of International Customer Success" in London marks a named leader for international accounts — a role that didn't exist here before. Two Staff+ engineering roles ("Staff+ Software Engineer, Claude Science" and "Staff+ Software Engineer, Enterprise AI Products") keep the technical bench moving, but the overall net is negative, consistent with a phase of trimming lower-priority lines.

Google DeepMind added 3 and closed 3. The "Specialist Humanoid Tele-Operator" in Mountain View is the single most signal-dense posting of the day — its job is to physically operate humanoid robots, producing demonstrations that feed model training. A "Senior Research Scientist, Frontier AI" in Toronto extends the lab's research nodes into Canada.

Meta AI added 8 over the day (21 over seven days). The four "Forward Deployed Engineer, Enterprise AI" roles in São Paulo and Singapore are the lead story: these are embedded commercial engineers who sit at customer sites and drive adoption, a model Meta is now scaling geographically. "Product Operations Manager, Meta Superintelligence Labs" in Menlo Park shows the superintelligence unit is building dedicated operations capacity. "Research Engineer, Fundamental AI Research (FAIR) - Generative Models/LLM Acceleration" and "AI Research Scientist," both in Tel Aviv, continue the deepening of FAIR's Israeli research presence.

NVIDIA posted 7 today (40 over seven days). The "Director, System Software Engineering - Metropolis Accelerated and Inferencing Software" role consolidates leadership over NVIDIA's video-AI inference platform. Three New College Grad 2026 postings cover AI DevTech, public-sector DevTech, and "ASIC Physical Design and Timing Engineer" — fresh pipeline for both software and silicon.

Cerebras added 4 with no closures. "Director/Sr. Manager, AI Inference Model Scaling" is the most senior of the batch and directly targets the inference throughput narrative that defines Cerebras's competitive positioning. Two new-grad roles (Kernel Engineer and ML Software Engineer - Integration & Quality) build out the engineering base.

Cursor opened 3 roles: "Forward Deployed Strategist" (Solutions), "GTM Finance - Sales Business Partner" (Finance), and "RVP, Digital CSM" (Customer Success). A regional VP of Customer Success signals that the customer base has grown large enough to require tier-based management — a maturation step for an AI coding tool that until recently grew mostly through product-led adoption.

MiniMax (Hailuo) added 2 roles in Shanghai: "科技合作与生态" (technology partnerships and ecosystem) and "政府关系 - 监管测" (government relations, regulatory track). A standalone regulatory-relations hire is uncommon among tracked Chinese labs and reflects the tightening domestic compliance environment.

Baidu AI, 01.AI (Yi), Moonshot AI (月之暗面 Kimi), Cohere, Scale AI, Apple AIML, SpaceXAI, and Perplexity each recorded one to four additions with limited new-job disclosure. Mistral AI, Lovable, Together AI, Pika Labs, Aleph Alpha, Genmo, Zhipu AI (智谱 AI), StepFun (阶跃星辰), Alibaba Tongyi Qwen (阿里通义千问), Tencent Hunyuan (腾讯混元), and SenseTime SenseNova (商汤日日新) posted no new roles today.

What it signals

The robotics cluster — DeepMind's tele-operator, NVIDIA's applied robotics engineer, OpenAI's robotics counsel — appearing on the same day is not coincidence in the usual sense; each company operates on its own hiring cadence. But their simultaneous visibility reflects a broader shift: embodied AI is crossing from pure research into infrastructure-building. DeepMind needs humans to run robots so algorithms can learn from physical demonstrations. OpenAI needs a lawyer who understands robotics commercial deals. NVIDIA needs engineers to ship the software that makes its chips useful on wheeled and legged platforms. The common thread is that each of these roles would have been premature two years ago.

The week's hiring leader, Microsoft AI, presents a different picture. Its 183 seven-day additions span at least four distinct vectors: the MAI Superintelligence Team in European research hubs, vector search infrastructure in Redmond, enterprise vertical sales in energy (oil and gas, Houston), and commercial leadership in emerging markets (South Africa). This breadth suggests a company that has identified AI as the primary growth lever and is staffing all of its facets in parallel rather than sequencing them. The closure rate (43 today) is high but proportional, indicating ongoing reprioritization rather than retrenchment.

The contrast with ByteDance Doubao/Seed is notable. Its 68 single-day closures — the largest of any company in today's data — stand against only 3 new postings, all in safety and evaluation. Whether the closures reflect consolidation of duplicate listings, a strategic pullback from certain research directions, or routine hygiene is unclear from the data alone. What is clear is that the new openings ByteDance chose to make public are precisely the functions under regulatory scrutiny: risk governance, model evaluation, and content safety for TikTok's engineering org. Chinese labs more broadly show a pattern today: MiniMax's regulatory-affairs hire, Baidu's "AI安全护栏-Go/Java开发工程师" (AI safety guardrail engineer), and ByteDance's safety evaluation cluster all point to compliance infrastructure being treated as a first-class engineering investment, not an afterthought.

The enterprise go-to-market expansion — OpenAI's three-timezone AWS seller push, Meta's emerging-market forward-deployed engineers, Anthropic's London customer success lead, Cursor's regional VP hire — collectively describes an industry shifting from capability demonstration to revenue execution. The labs that spent 2024 and early 2025 proving what their models could do are now staffing the motions required to turn that capability into contracted, recurring enterprise revenue. That transition is visible in job titles today across at least five companies simultaneously.