AI Hiring Brief · 2026-08-30

32 frontier AI labs · 47 roles added in 24h · agihunt.info/en/jobs

Today at a glance

Silicon Valley's biggest single-day churn: Microsoft AI added 20 roles but closed 40 — the sharpest daily contraction in today's data — even as it keeps the largest 7-day addition count (188) of any company tracked.
ByteDance's pullback continues: Doubao/Seed (字节跳动豆包/Seed) closed 25 roles today with zero additions, a second straight heavy-closure day, though its 937 open roles remain the single largest count in the dataset.
Safety and policy hiring keeps pace with sales: Anthropic posted a Safeguards Enforcement Lead, Cyber Harms and a Regulatory Counsel, AI Regulation, US alongside a Strategic Account Executive, Tech — trust-and-safety and commercial hiring landing side by side.
A China lab goes vertical: ModelBest (面壁智能) posted two Beijing-based roles explicitly targeting automotive — 多模态算法Leader-汽车行业 (Multimodal Algorithm Lead, Automotive) and 智能座舱多模态大模型算法leader (Smart Cockpit Multimodal LLM Algorithm Lead) — a specific in-car AI push rather than general hiring.
Most of China's leading labs sat still today: Zhipu AI, Baidu AI, Tencent Hunyuan, MiniMax, Alibaba Tongyi Qwen, SenseTime SenseNova and DeepSeek all show zero net change on the day.

Company moves

OpenAI (758 open; 63 added over 7 days, 6 today against 4 closures) mixed enterprise-partner GTM — a Partner Director tied to PwC spanning San Francisco, New York and Seattle — with legal (Senior Counsel, AI & Product Compliance) and a Safety Response Operations Lead in User Operations.

Anthropic (571 open; 84 added over 7 days — the fastest pace among global labs tracked — 5 today against 3 closures) paired trust-and-safety hires (Safeguards Enforcement Lead, Cyber Harms; Product Manager, Multi-Cloud Trust & Safety) with revenue-side build-out: Strategic Account Executive, Tech in Seattle, Regulatory Counsel, AI Regulation, US, and Director, Finance Systems - Revenue Systems Engineering.

Google DeepMind (73 open; 14 added over 7 days, 1 today) posted a single Senior Director, Product, Model Services and Consumer Verticals role in Mountain View — a consumer-product-management signal.

Microsoft AI (602 open; 188 added over 7 days, 20 today against 40 closures) ranged from Principal Software Engineer and Software Engineer II seats in Redmond to a Cloud and AI Platforms (CAIP) Specialist in Zürich, matching Solution Specialist - AI Workforce and Solution Area Specialists-AI Workforce roles in São Paulo, and an Applied Scientist, AI Economics (TokenOps, FinOps) — broad engineering and international specialist-sales activity even as closures ran two-to-one against additions.

Meta AI (152 open; 13 added over 7 days, 1 today against 3 closures) added an AI Research Scientist, Video Generation and Post Training seat inside FAIR, spanning Menlo Park, Seattle and New York.

NVIDIA (575 open; 50 added over 7 days, 2 today against 1 closure) posted a Director, Engineering - Autonomous Vehicles and a System Level Product Development Engineer, both in Santa Clara.

Apple AIML (285 open; 15 added over 7 days, 5 today against 1 closure) clustered around on-device inference and speech: On-device ML Performance Engineer and On-device ML Engineering Manager (Tools & Services) in Seattle, an On-Device ML Infrastructure Engineer (CoreML Runtime) and a second Performance Engineer role in Cupertino, plus a Speech Processing ML Algorithm Engineer.

SpaceXAI (252 open; 15 added over 7 days, 2 today against 2 closures) added a Controls Technician and a Journeyman Electrician, both under the Data Center org and both spanning Memphis, TN and Southaven, MS — continued physical data-center buildout.

Perplexity (97 open; 6 added over 7 days, 0 today against 2 closures) had no new postings today, only closures.

Together AI (62 open; 2 added over 7 days, 1 today) added a GTM Data Analytics Engineer in Revenue Operations, San Francisco.

Scale AI (219 open; 15 added over 7 days, 1 today) added an Operations Associate in Gen AI Operations based in Argentina, extending its international data-operations footprint.

Cerebras (109 open; 4 added over 7 days, 1 today) posted a VP Legal, Strategic Transactions role in Sunnyvale.

ModelBest (面壁智能; 138 open; 12 added over 7 days, 2 today) posted both new roles into automotive multimodal AI, as noted above — 多模态算法Leader-汽车行业 and 智能座舱多模态大模型算法leader, both in Beijing.

Meituan LongCat (美团·龙猫; 72 open) had a pure contraction day — 4 closed, none added.

ByteDance Doubao/Seed (字节跳动豆包/Seed; 937 open, still the largest count tracked; 7 added over 7 days, 0 today against 25 closed) posted a second consecutive heavy-closure day, following 61 closures the day prior, even as its open-role total remains the highest in the dataset.

Mistral AI, Cursor, Lovable, Cohere, Pika Labs, Aleph Alpha, Genmo, Moonshot AI (月之暗面 Kimi), DeepSeek (深度求索), Zhipu AI (智谱 AI), StepFun (阶跃星辰), Baidu AI (百度 AI), Tencent Hunyuan (腾讯混元), MiniMax (Hailuo) (MiniMax 海螺), 01.AI (Yi) (零一万物), SenseTime SenseNova (商汤日日新), and Alibaba Tongyi Qwen (阿里通义千问) posted no changes today.

What it signals

The gap in hiring velocity between global and China-based labs is stark in the 7-day figures. Anthropic (84 added) and Microsoft AI (188 added) are moving far faster than any China lab in the dataset — the closest China figure is StepFun's 30 over 7 days, followed by Tencent Hunyuan (15), Baidu AI (16) and ModelBest (12), with several major names — Zhipu AI, Alibaba Tongyi Qwen, SenseTime SenseNova — flat at zero net change for the full week. That doesn't necessarily mean China labs are hiring less overall, but on this window their open-role counts are simply not turning over the way OpenAI's, Anthropic's or Microsoft's are.

Closures are the other half of the story. Microsoft's 40 closures today against 20 additions, and ByteDance's 25 closures against zero additions — its second heavy-closure day in a row — both point to postings churn rather than straightforward net growth. Large employers with hundreds of open roles (Microsoft at 602, ByteDance at 937) appear to be actively pruning and refreshing requisitions, not just stacking new ones on top of old.

Where roles are new, the content skews toward two recurring themes at the frontier labs: trust-and-safety/policy hiring (Anthropic's Safeguards Enforcement Lead and Regulatory Counsel; OpenAI's Safety Response Operations Lead) running in parallel with commercial expansion (Anthropic's Strategic Account Executive; Scale AI's Argentina operations role; Together AI's GTM analytics hire). That pairing suggests labs scaling governance functions and revenue functions at the same time as enterprise adoption grows, rather than sequencing one after the other.

The one clear vertical signal among China's leading labs today is ModelBest's pair of automotive-focused multimodal roles in Beijing — a specific bet on in-car AI rather than general model or infrastructure hiring, and notable precisely because it stands out against an otherwise quiet day across the rest of China's tracked labs.