AI Hiring Brief · 2026-08-29

32 frontier AI labs · 102 roles added in 24h · agihunt.info/en/jobs

AI Hiring Brief · 2026-08-29

32 frontier AI labs tracked · 102 roles added in the last 24 hours

Today at a glance

Microsoft doubles down on silicon: Director, Silicon Design and Principal Memory Subsystem Engineer (Design or DV) postings across Raleigh, Mountain View and Redmond point to continued custom-chip work inside the Silicon Engineering org, alongside a raft of Copilot- and CoreAI-branded software roles.
Anthropic builds out physical infrastructure and safety policy in parallel: Engineering Manager, Hardware Platform Security, Incident Management Lead, Data Center Security, and NPI Sourcing Manager, Data Center Power & Cooling landed the same day as Head of Policy Design, Societal Harms — data-center hardening and trust-and-safety governance advancing together.
OpenAI extends its APAC/ANZ sales footprint: new Account Director openings in Tokyo and Sydney, an Account Director, Government role covering Australia, and a Solutions Engineering, Ads Solutions (Singapore) posting all point toward deeper enterprise and government go-to-market coverage outside the US.
ByteDance Doubao/Seed posts the day's sharpest contraction: 61 roles closed against just 2 added — the largest single-day closure count in today's data, at a company that still carries the single largest open-role count tracked (911).
StepFun (阶跃星辰) leads the pace among China-based labs: 30 roles added over the past 7 days, ahead of every other China lab tracked, with a Senior AI Application QA Engineer (AI应用高级测试工程师) posted today.

Company moves

OpenAI (756 open; 59 added over 7 days, 13 today against 5 closures) mixed international commercial roles — the Tokyo, Sydney and Singapore postings above, plus Account Director, Commercial in New York — with product and support hires: Software Engineer, Astral in the Applied AI org, and a Manager, Senior Support Engineering seat in Dublin.

Anthropic (568 open; 87 added over 7 days, the fastest pace of any global lab, 15 today against 7 closures) paired the hardware-security and data-center cluster above with commercial expansion — Reseller Enablement Lead and Finance & Strategy GTM, EMEA in London, plus Reseller Operations Manager, Japan & Korea and an Enterprise Account Executive covering Retail/CPG/Trading in Tokyo — and a Senior Manager, Compute Accounting role, suggesting compute cost governance is scaling alongside the infrastructure build-out.

Google DeepMind (72 open; 13 added over 7 days, 2 today against 5 closures) posted a Director, Trust and Safety, Trusted Experiences role spanning seven US cities and a Senior Mechanical Engineer, Plastics and Polymers in Cambridge, MA — a small but notable materials-engineering signal alongside safety hiring.

Microsoft AI (612 open; 180 added over 7 days — the largest 7-day count in the dataset — 43 today against 51 closures, also the day's largest closure count) combined the silicon-design roles above with a wide APAC/EMEA solution-architecture and sales push (Oslo, Tokyo, Taipei, Bangkok-adjacent Mexico City, Vancouver) and several Principal/Senior Software Engineer openings in Redmond tied to CoreAI and Compute Graph Compiler work.

Meta AI (154 open; 12 added over 7 days, 3 today against 4 closures) added Research Scientist, Machine Learning in Sunnyvale and two Systems ML infrastructure engineering roles split between Toronto and Menlo Park — continuing to staff foundational ML infrastructure.

NVIDIA (英伟达, 574 open; 51 added over 7 days, 3 today against 7 closures) added Senior System Test Engineer, Senior Solution Architect for Sales AI Applications, and Senior Applied Research Scientist, Data Curation — spanning hardware validation, GTM and data-quality work for model training.

SpaceXAI (252 open; 13 added over 7 days, 6 today against 1 closure) posted a cluster of data-center trades roles in Memphis and Southaven — Controls Technician, Plumbing Technician, Fire Protection Technician, Network Connectivity Engineer, Supervisor IT Support and Sous Chef — physical build-out roles for an active data-center campus.

Perplexity (99 open; 6 added over 7 days, 2 today, 1 closure) added Director, Litigation and Public Policy Associate in Washington, D.C., both in Legal — a policy-facing hire pattern as the company scales its DC presence.

Cursor and Lovable each added a single role today (Sales Development Representative in New York for Cursor; AI Scaled Customer Success Programs in Boston for Lovable), both sales/CS hires rather than engineering.

Cerebras (108 open; 3 added over 7 days, 2 today against 3 closures) posted Senior Supplier Quality Engineer and Senior Data Engineer roles in its Hardware org, both in Sunnyvale.

Among China-based labs, Moonshot AI (月之暗面 Kimi, 107 open; 6 added over 7 days) added a Growth Creative role (增长素材创意) and a Global HR opening in Haidian. DeepSeek (深度求索, 37 open) added one R&D Infrastructure Engineer (研发基础设施工程师) posting across Gongshu and Haidian. StepFun (阶跃星辰, 179 open) posted the QA role noted above and led China's 7-day pace at 30. ModelBest (面壁智能, 136 open; 10 added over 7 days) added three Beijing-based roles — Test Development Engineer for its legal-AI business (测试开发工程师-法律智能业务), an LLM Data Engineer/Data Pipeline role, and LLM Data Operations (大模型数据运营). MiniMax (MiniMax 海螺, 197 open) added an LLM Systems Architect role with a storage/database background (大模型系统架构师，存储/数据库背景) across Beijing and Shanghai. ByteDance Doubao/Seed (字节跳动豆包/Seed, 911 open) added two Coding/Agent evaluation product roles for its Volcano Ark platform even as it closed 61 roles today, by far the largest closure round tracked.

The remaining labs were flat today: Mistral AI, Pika Labs, Aleph Alpha, Genmo, Cohere, Together AI, Scale AI, Zhipu AI, Meituan LongCat, Baidu AI, Tencent Hunyuan, 01.AI (Yi), SenseTime SenseNova and Alibaba Tongyi Qwen posted no new roles today, and Apple AIML added just one (ISE, System Experience — Senior Software Engineer, macOS System Experience).

What it signals

Microsoft AI's 180 additions over the past 7 days is the largest of any lab in this dataset, but today's own numbers — 43 added against 51 closed — show that raw volume is running alongside heavy churn rather than pure net growth; the closure count today is the highest of any company tracked. Anthropic offers the cleaner growth signal: 87 net additions over 7 days against just 7 closures today, with hiring spread deliberately across hardware security, data-center power infrastructure, compute-cost accounting, and international reseller/GTM roles in London and Tokyo. That combination — physical infrastructure, financial controls, and commercial expansion moving together — suggests Anthropic is staffing for a compute build-out it also intends to monetize and govern, not just scale.

OpenAI's 59 additions over 7 days read as broader and more evenly distributed: today's postings touch marketing, ads support, recruiting operations, and enterprise sales across San Francisco, New York, Tokyo, Dublin, Singapore and Sydney, without a single dominant function. That breadth, paired with a modest 5 closures today, points to steady multi-front growth rather than a reorganization.

The gap between global and China-based labs remains wide in absolute terms. StepFun's 30 additions over 7 days is the strongest pace among China labs, but most others — Zhipu AI, Alibaba Tongyi Qwen, SenseTime SenseNova — posted zero new roles today and modest or no 7-day movement. ByteDance Doubao/Seed is the outlier: it holds the largest open-role count in the entire dataset (911) yet added only 7 roles over the past week while closing 61 in a single day, a contraction pattern distinct from anything else tracked. Baidu AI and Tencent Hunyuan (16 and 15 added over 7 days, respectively, with zero today) show slower, steadier movement rather than the sharp swings seen at ByteDance or Microsoft.

Read together, the hardware and safety threads are the clearest cross-company pattern today: Anthropic's hardware-platform-security and data-center roles, Microsoft's silicon-design pair, and NVIDIA's hardware-validation and data-curation hires all point to labs continuing to invest in physical compute capacity even as commercial and support functions expand around them — infrastructure and go-to-market are not being staffed sequentially, but at the same time.