AI Hiring Brief · 2026-07-16

32 frontier AI labs · 149 roles added in 24h · agihunt.info/en/jobs

Today at a glance

Red-team hiring standardizes across three labs simultaneously: Anthropic posted "Lead, Frontier Red Team (Cyber)," Apple AIML posted "AIML - Machine Learning Engineer, Red Teaming - Responsible AI and Safety," and OpenAI posted "Research Engineer, Privacy" on the same day — covering model, device, and application layers in a single session, a pattern more consistent with shared external pressure than independent internal planning.
Anthropic's broadest single-day expansion on record: Twelve new roles spanning safety enforcement, data center architecture, capacity and data infrastructure engineering, global sales leadership, enterprise product, public-sector sales, investor relations, and a London applied-AI seat — with "Director, Investor Relations" and "Global Head of Sales & Partner Enablement" arriving together as a textbook fundraise-and-commercialize one-two signal.
Cohere assembles a global communications apparatus: PR Directors for APAC (Singapore/Korea) and EMEA (UK/Belgium/France/Germany), plus an executive communications manager, product communications manager, coordinator, and a US government affairs lead all posted the same day — alongside Saudi Arabia and Japan sales roles — the clearest pre-raise or pre-listing brand-infrastructure pattern in today's data.
Scale AI formalizes Agent Oversight as a staffing category: "Senior Machine Learning Engineer, Agent Oversight" and "Senior Software Engineer, Agent Oversight" posted together in San Francisco and New York — agent governance has crossed from research topic to dedicated engineering headcount.
NVIDIA moves deeper into robotics and automotive silicon: "Principal SoC Architect, Robotics and Automotive," "Principal Technical Program Manager - Autonomous Vehicles," and "Software Engineer, AI Networking Architect" all landed today, indicating hardware customization is pushing beyond data center into application-specific end markets.

Company moves

Anthropic added 12 roles today (58 over seven days) across safety ("Lead, Frontier Red Team (Cyber)," "Software Security Engineering Manager, Secure Frameworks," "Safeguards Enforcement Analyst, Violence & Extremism"), compute infrastructure (two Staff+ Software Engineers and a "Data Center Architect, CSA"), commercialization ("Global Head of Sales & Partner Enablement," "Product Manager, Enterprise," "Enterprise Account Executive, State & Local Sales"), and a "Director, Investor Relations." A London "Applied AI Engineer, Beneficial Deployments" extends the geographic footprint. The investor relations and global sales leadership appearing in the same daily batch is the sharpest signal this tracker has logged for a near-term Anthropic fundraise.

Microsoft AI posted 39 additions against 55 closures today (154 added over seven days). The "AI Workforce" commercial line is spreading internationally: "Partner Solution Architect: AI Workforce" in Bangkok and Singapore, "Digital Enterprise Solution Sales Specialist - AI Workforce - Polish speaking" in Dublin. "Principal Privacy Product Manager, Health AI" signals a healthcare vertical taking shape. Research Science internship openings in Beijing and Shanghai suggest the China research presence remains active.

Cohere added 8 roles today (138 over seven days). Five are communications hires — "PR Director, APAC," "PR Director, EMEA," "Product & Technology Communications Manager," "Senior Manager, Executive Communications," "Corporate Communications Coordinator" — alongside a "Lead - US Government Affairs & Public Policy" in Washington, DC, a "Senior Account Executive - KSA (Eastern Province)" in Saudi Arabia, and a "Partner Development Manager - Japan" in Tokyo. The geographic and functional breadth across a single day reads as deliberate brand infrastructure construction before a step-change in public profile.

ByteDance Doubao/Seed (字节跳动豆包/Seed) added 37 roles today — 977 over seven days, a scale no other company in this dataset approaches. Today's batch includes 3D generation researchers ("3D生成算法工程师-Seed," a 3D generation algorithm engineer; "多模态世界模型算法研究员-Seed," a multimodal world-model researcher), three medical AI roles under the Xiaohe Health brand covering LLM evaluation, data security operations, and application safety operations, TikTok content-safety platform LLM engineers, and a code-intelligence specialist for TRAE ("大模型算法专家（代码智能方向）-TRAE"). Agent roles appear across Seed, CapCut, and Doubao Mobile Assistant simultaneously — broad penetration across product lines rather than a concentrated push on any one.

Scale AI added 5 roles today (187 over seven days). The Agent Oversight pair is the headline: dedicated ML and software engineering headcount for monitoring agents in production marks the function as operational, not exploratory. Three public-sector roles — including a "Forward Deployed Product Manager, Public Sector" in DC — confirm the government contract build continues.

NVIDIA added 10 roles today (48 over seven days), with silicon work dominant: "Principal SoC Architect, Robotics and Automotive," "Senior SOC Design Engineer" in Bengaluru, and multiple ASIC design and verification engineers form a coherent hardware design cluster alongside "Software Engineer, AI Networking Architect," suggesting AI networking fabric is now a dedicated internal capability.

Apple AIML added 6 roles today (21 over seven days). "ML Agent Engineer" in the Bay Area is a rare explicit agent title at Apple. "Research Scientist - LLM Efficiency" in Zurich continues the European research build. The red-teaming safety role aligns Apple with the cross-industry pattern visible at Anthropic and OpenAI today.

OpenAI added 6 roles today (72 over seven days). "Founding Data Scientist, Pricing & Monetization" — the "founding" qualifier signals monetization analytics is being built from scratch. "Technical Deployment Lead, Semiconductors" in Forward Deployed Engineering suggests custom chip deployments are moving to customer-facing phases. "Engineering Manager, Model Flywheel" treats training-data flywheel management as its own leadership function.

Cerebras added 6 roles today, bringing its seven-day total to 104 — a full open-position list rebuilt in a week. Cluster test engineering, simulation, co-design, ML systems integration, and DevOps, alongside a "Technical Business Development Manager" in Hardware Departments, indicate a company scaling both the engineering base and the hardware sales motion together.

Cursor posted 3 roles today: "RVP, Customer Success, Strategic & Geo Enterprise (AMER)," "Engineering Manager, Agent & Product Security," and "Public Sector Sales Strategy & Operations Manager." Enterprise retention, agent security, and government market entry arriving simultaneously is a clear upmarket pivot signal.

Google DeepMind added 4 roles today (34 over seven days). "Research Scientist, AnthroKrishi" in Bengaluru is a dedicated agricultural AI project role — an unusual vertical extension into physical-world applications. Gemini App productization continues with data engineering and UX program management hires.

StepFun (阶跃星辰) posted one "音频研发工程师" (audio R&D engineer). Baidu AI (百度 AI) added "数字人FDE（J102247）" (a forward-deployed engineer for the digital-human product). 01.AI (零一万物) posted two internship roles including a "塞尔维亚语实习生" (Serbian-language intern), hinting at Balkan market exploration. Mistral AI, DeepSeek (深度求索), Zhipu AI (智谱 AI), Alibaba Tongyi Qwen (阿里通义千问), Tencent Hunyuan (腾讯混元), MiniMax (MiniMax 海螺), Meituan LongCat (美团·龙猫), ModelBest (面壁智能), SenseTime SenseNova (商汤日日新), Pika Labs, Aleph Alpha, Lovable, and Genmo posted no new roles today.

What it signals

The seven-day figures reveal a stark velocity bifurcation. ByteDance Doubao/Seed's 977 additions over seven days is not statistical noise — it reflects a structural difference in operational scale, with every other company in this dataset more than five times smaller by this measure. Scale AI (187), Cohere (138), and Cerebras (104) represent a second tier undergoing what appear to be deliberate build-out cycles: not routine backfill but wholesale expansion of new functions. That three companies of different business models — data labeling and evaluation, enterprise LLM APIs, and AI hardware — are all in simultaneous large-scale hiring suggests the industry's investment phase is broadening rather than concentrating.

The simultaneous red-team postings across Anthropic, Apple AIML, and OpenAI point to a shared external stimulus. Labs that are planning independently of one another rarely synchronize safety engineering hires at this granularity. Regulatory deadlines, enterprise procurement requirements, or a notable recent incident are the more likely drivers than coincidence. The fact that the roles span model safety (Anthropic's cyber red team), device-layer responsible AI (Apple), and application-layer privacy (OpenAI) suggests comprehensive coverage is being demanded — not just marquee gestures.

Cohere's communications and government-affairs build — five PR roles plus regional leads for Saudi Arabia, Japan, and US government in a single day, against a 138-role seven-day backdrop — is the most legible near-term-event hiring signal in today's data. Infrastructure of this kind is not built speculatively; it is built when the communications demand is already visible on the horizon.

The chip-layer investment at NVIDIA, OpenAI, and Cerebras, taken together, suggests silicon is becoming a differentiated competitive front beyond data center throughput. NVIDIA is building dedicated SoC architecture for robotics and automotive at the principal level. OpenAI is deploying semiconductor specialists to customer sites. Cerebras rebuilt its entire open-position list in a week. The gap between companies investing in hardware capability and those that remain software-only is widening, and today's postings suggest that gap is being treated as a strategic concern at multiple points in the stack simultaneously.