AI Hiring Brief · 2026-08-18

32 frontier AI labs · 91 roles added in 24h · agihunt.info/en/jobs

Today at a glance

Microsoft AI's health cluster stands out: Microsoft AI added 15 roles today and 206 over the past 7 days — the largest tally of any company tracked — with a visible Health AI cluster in London (Member of Technical Staff - Software Engineer, Health AI; Product Designer - MAI Health; Member of Product Staff - Health AI), even as 31 roles closed the same day.
Anthropic pairs international sales growth with security build-out: Anthropic added 20 roles today, the most of any company, split between customer-facing GTM hires in Munich, Paris, Dublin and Sydney (Forward Deployed Engineer, Customer Success Manager, Enterprise Account Executive) and senior security leadership (Head of Enterprise Risk, Regional Safety & Security Manager, Americas, Physical Security Design Engineer).
Agent infrastructure hiring shows up on both sides of the Pacific: OpenAI posted Product Engineer, Full Stack - Agents in its Applied AI org, while StepFun (阶跃星辰) listed Agent Infra开发工程师/架构师 (Agent Infrastructure Engineer/Architect) — both labs staffing up the plumbing behind agent products.
ByteDance keeps cutting its pipeline: ByteDance Doubao/Seed (字节跳动豆包/Seed) closed 39 roles today, the steepest same-day closure count of any company tracked, against just 7 additions — even as its 914 open roles remain the highest count in the dataset.
China's frontier labs stay mostly still: Zhipu AI, Alibaba Tongyi Qwen, Mistral, Cerebras, Moonshot AI (Kimi) and DeepSeek all logged zero net movement today, and combined 7-day additions across China's 13 tracked labs (72) remain far below Microsoft AI's own 7-day total (206), among the 32 companies tracked overall.

Company moves

OpenAI: Added 8 roles today, leaning toward safety operations and go-to-market — Senior Operations Analyst, User Safety & Risk Operations; Data Scientist, Cybersecurity; Strategic Partner Manager, Monetization Partnerships; Account Director, Startups in Munich — plus the agents-focused Product Engineer, Full Stack - Agents, against 16 closures.

Anthropic: Its 20 additions split between international sales/customer-success roles (Munich, Paris, Dublin, Sydney) and security leadership hires, plus Software Engineer, Research Data Platform and Head of Treasury Strategy & Transformation — the week's second-largest addition total in the dataset at 92.

Google DeepMind (谷歌 DeepMind): Added an Engineering Technician, Robotics, DeepMind role in Indiana alongside a Financial Analyst, GenAI and a Science and Strategic Initiatives software engineer — but 10 closures today outweighed the 3 additions.

Microsoft AI (微软 AI): The Health AI cluster in London leads today's postings, alongside sales specialist roles in Auckland and Malaysia and a data-center program manager spanning Dublin, Amsterdam, Milan and London — 31 closures accompanied the 15 additions.

Meta AI: Two roles — Engineering Manager, Enterprise Solutions - Latin America in São Paulo, and Production Systems Engineer, AI Systems in AI Infrastructure — a small international enterprise push alongside steady infra hiring.

NVIDIA (英伟达): Chip-design roles continue — Senior DFT Engineer and Senior Signal and Power Integrity Engineer in Santa Clara — alongside a Relational Foundation Model Engineer for its Modern Data Stack team.

Apple AIML (苹果 AIML): Breadth across the ML org — a Responsible AI and Safety operations program manager, an Accessibility applied research scientist, and a Health AIML program manager — spanning safety, accessibility and health rather than one focus area.

SpaceXAI: Both new roles sit in Safety — Program Manager, Prohibited & Regulated Content and Program Manager, Harmful Activity — a trust-and-safety-specific push.

Perplexity: Added two legal hires, Principal Counsel, Corporate and Litigation Paralegal, pointing to growing legal/compliance overhead as the company scales.

Cursor: Added a Sales Development Manager - EMEA in London, a small step toward international sales coverage.

Lovable: Added a Social Media Lead in Stockholm as its marketing function builds out.

Cohere: Added Member of Technical Staff - RL Environments in London and a Senior Account Executive covering Montreal/Atlantic Canada — research and sales moving in parallel.

Together AI: Added Senior Product Manager, Model APIs & Developer Experience and a Strategic Finance Senior Associate - Compute, reflecting its infrastructure-as-a-service model.

Scale AI: No new detailed postings today, just 2 closures, though its 7-day tally of 12 shows hiring continuing beneath the surface.

StepFun (阶跃星辰): Posted Agent Infra开发工程师/架构师 (Agent Infrastructure Engineer/Architect) and 测试开发工程师 (Test Development Engineer) across Xuhui and Haidian, plus an indirect-procurement role — infra and QA hiring pointing to agent product build-out.

ModelBest (面壁智能): Closed 6 roles today with no new postings visible — flat for the week on a net basis.

Meituan LongCat (美团·龙猫): Added a Technical Project Manager (PMO) for its Longcat model, spanning Beijing, Shanghai and Shenzhen.

Baidu AI (百度 AI): Added a Token Hub resource-operations role, an AI Product Manager, and an anti-intrusion security engineer — spanning commercialization infrastructure and platform security.

ByteDance Doubao/Seed (字节跳动豆包/Seed): Closed 39 roles today while adding 7 — several new postings are safety-related, including 风险应急响应Leader-大模型安全群组 (Risk Emergency Response Lead, Large Model Safety Group) and a training/inference stability role under its TikTok content-safety platform — safety hiring persisting even as posting volume shrinks.

Tencent Hunyuan (腾讯混元): Added an evaluation-infrastructure observability engineer in Beijing/Shenzhen against 3 closures.

MiniMax (Hailuo, 海螺): Added an Open Platform backend engineer role, with 3 closures the same day.

01.AI (Yi, 零一万物): Added an AI consulting/FDE specialist role plus UX and marketing internships — enterprise deployment and entry-level hiring side by side.

Mistral, Cerebras, Pika Labs, Aleph Alpha, Genmo, Moonshot AI (Kimi), DeepSeek, Zhipu AI, SenseTime SenseNova (商汤日日新) and Alibaba Tongyi Qwen logged no net movement today.

What it signals

The gap between Microsoft AI and everyone else keeps widening. Its 206 net additions over the past 7 days are nearly three times the combined 72 added across China's 13 tracked labs, and larger than the 7-day total of any other single company by a wide margin. That volume, concentrated partly around a new Health AI cluster, suggests a deliberate vertical push into healthcare applications layered on top of already-massive general hiring — not just broad-based scaling.

Anthropic's pattern this week reads differently: its additions cluster around enterprise go-to-market expansion into new geographies (Munich, Paris, Dublin, Sydney) happening in the same window as senior security and enterprise-risk leadership hires. Scaling international revenue and building out physical/regulatory risk functions in parallel, rather than sequentially, points to a lab treating trust infrastructure as a prerequisite for expansion rather than an afterthought.

Safety and trust-related roles surfaced across multiple labs today — SpaceXAI's two Safety program manager postings, ByteDance's large-model safety and content-governance hires, and Apple's Responsible AI and Safety program manager — independent of each company's core model work. That breadth suggests safety/trust hiring is becoming a standing function across the industry rather than a reactive one tied to a single incident or launch.

Among China's labs, the picture is uneven rather than uniformly slow. Baidu AI and StepFun kept adding at a modest but steady pace (20 and 11 over 7 days respectively), while Zhipu AI and Alibaba Tongyi Qwen showed zero net movement and ByteDance ran its steepest closure day in the dataset. That split — some labs still filling roles while the largest, ByteDance, visibly prunes its pipeline — looks more like selective consolidation than a broad pullback in China's AI hiring.