Late on June 25, DeepSeek posted job openings across several Chinese social-media platforms and said, in plain terms, that it is “striving to at least double the size of all departments.” It is the most aggressive expansion the famously lean Hangzhou startup has ever attempted — 33 roles across seven categories, based in Beijing and Hangzhou, with interns welcome on every team.
The timing is no accident. The hiring push lands right after DeepSeek closed its first major outside funding round — a reported ~50 billion yuan (about $7.4 billion) that, per people familiar with the deal, values the company north of $50 billion. By Friday afternoon a company recruiter said the late-night post had already drawn more than 10,000 résumés.
What’s open
DeepSeek grouped the 33 positions into seven buckets. The shape of the list is the story: this is no longer a pure research lab hiring a handful of algorithm specialists, but a full-stack AI company staffing the entire pipeline from supercomputing clusters to product and compliance.
| Category | What it signals | Notable roles |
|---|---|---|
| Full-stack development & algorithms | Largest block | Server-side dev, pre-training data engineer, AI search algorithm & architecture, Agent Harness team |
| AI core system R&D | Scaling the compute stack | Supercomputing-cluster R&D, high-performance operators/comms/compiler, training & inference frameworks |
| Deep-learning research | Frontier bets | Researchers on continual learning and self-evolving systems |
| Model data strategy (PM & engineering) | Data as a product | Agent-data PMs; specialised-domain data PMs in non-English languages, medicine and law |
| Operations | Reliability at scale | AI compute-cluster performance & platform operations |
| Product | Turning models into apps | AI product managers and product operations |
| Corporate functions | Becoming a real company | Administrative and support roles |
One detail stands out for builders: the Agent Harness team. DeepSeek has previously described a “Model + Harness = Agent” framework, and reporting ties this hiring to AI-coding tools meant to compete with products like Anthropic’s Claude Code — an end-to-end coding agent that lives in developers’ daily workflow.
Why now
For two years DeepSeek’s brand was leanness — frontier reasoning models trained at a fraction of Silicon Valley’s cost, shipped by a small, secretive team. This move reads as a deliberate break from that “lean AI” identity. The funding round, reportedly anchored by a large personal injection from founder Liang Wenfeng alongside strategic backers, gives the company the balance sheet to convert capital into headcount fast.
Crucially, DeepSeek chose to grow every department rather than just engineering. That suggests management sees its bottleneck not as raw research talent but as organizational infrastructure — product, operations, sales motion and legal/compliance depth needed to ship and support real products at scale.
The framing is unusually grand. “Humanity now stands on the eve of AGI,” the announcement said, pitching newcomers a “front-row seat” and promising that fresh hires will take on “the most core and important tasks directly.”
Why it matters
A few takeaways for anyone tracking the model landscape:
- The China frontier is consolidating around fewer, better-capitalized labs. A $50B+ valuation and a 10,000-résumé response signal that DeepSeek can now compete for talent on capital, not just mission — narrowing one of the West’s structural advantages.
- Agents and coding tools are the next battleground. The Agent Harness hire, plus pre-training-data and AI-search roles, point at a product roadmap aimed squarely at agentic coding and search — the same ground OpenAI’s Codex and Anthropic’s Claude Code are fighting over.
- Markets are watching. When the news broke, Nvidia shares slid as much as 3.5%, a reminder that DeepSeek’s cost-efficient approach still rattles assumptions about AI infrastructure spending.
If you want context on where DeepSeek sits among frontier models, see our profile of DeepSeek V4-Pro and the broader 2026 LLM landscape guide.
A note on the numbers: the funding figure (~50 billion yuan / ~$7.4B), the valuation and the list of backers come from reporting citing people familiar with the deal and have not been officially confirmed by DeepSeek. Treat them as well-sourced but provisional.
Sources
- South China Morning Post — DeepSeek hiring spree as it pursues AGI
- Caixin Global — DeepSeek Plans Major Hiring Spree After $7.4 Billion Funding Round
- The Edge / Bloomberg — DeepSeek plans to double staff across all departments
- TMTPost — DeepSeek Abandons ‘Lean AI’ Strategy as Funding Round Vaults Valuation Past $50 Billion