How to Apply to 01.AI

10 min read Last updated April 20, 2026

ResumeGeni's employer crawl shows 01.AI runs its own custom application flow. Standard parser rules still apply: conventional section headings, text bullets, no tables. See the general ATS formatting guide.

Key Takeaways

  • 01.AI is Kai-Fu Lee's bet on Chinese frontier AI, founded March 2023, valued around $1 billion, with the Yi model series and the Yi-Lightning closed-source frontier model as flagship products.
  • The company pivoted in early 2025 away from the pure foundation model race toward Sea AI for Asia-Pacific markets and commercial enterprise AI; this reframes what hiring success looks like.
  • DeepSeek's January 2025 V3 and R1 releases reset competitive expectations across the industry and pressured every Chinese frontier lab, 01.AI included.
  • Hiring is heavily craft-driven and demanding; the bar has not relaxed even as the strategy has narrowed.
  • Beijing and Shanghai roles favor Mandarin-fluent candidates; Singapore roles operate in English; bilingual candidates have a real edge.
  • GPU sourcing under US export controls is a real operational constraint, not a marketing footnote, and shapes infrastructure roles.
  • Compensation is competitive within Chinese AI but below US frontier labs; equity in a post-pivot startup carries genuine risk and possible upside.
  • Foreign candidates should expect one to three months of visa processing for Beijing roles and the standard Singapore Employment Pass timeline for Singapore roles.

Source basis: This guide combines the company's public careers materials, detected ATS-provider data, and ResumeGeni analysis. Employer-specific details should be read alongside the Sources section below; interview-culture guidance may synthesize public candidate reports when official documentation is limited.


About 01.AI

01.AI (零一万物, 01.AI Inc.) is a Chinese large language model startup founded in March 2023 by Kai-Fu Lee (李开复), the former president of Google China, ex-Microsoft Research Asia executive, and founder of the venture firm Sinovation Ventures. Headquartered in Beijing with additional engineering presence in Shanghai and a growing Singapore office for international expansion, the company has scaled to roughly 200 or more employees in under two years and reached unicorn status within eight months of founding after closing a Series A round in November 2023 led by Alibaba Cloud and Sinovation Ventures, with reported total funding above $300 million and a valuation around $1 billion. The company became visible through its Yi model series. Yi-34B, released in July 2023 under the permissive MIT license, briefly topped open-source leaderboards on Hugging Face and benchmarked competitively against Meta's Llama 2 70B at roughly half the parameter count. Subsequent releases include Yi-VL for vision-language tasks, the Yi-1.5 update in September 2024, and Yi-Lightning in October 2024, a closed-source production model that ranked sixth on the LMSYS Chatbot Arena at release, briefly placing it above Anthropic's Claude 3.5 Sonnet in human preference voting and signaling that Chinese frontier labs were closing the gap with US counterparts. In early 2025 Kai-Fu Lee publicly announced a strategic pivot. He acknowledged that 01.AI could no longer keep pace with the capital intensity of the foundation model race against Alibaba's Tongyi Qianwen (Qwen), ByteDance's Doubao, Baidu's ERNIE, and the Hangzhou-based DeepSeek, whose V3 and R1 releases in January 2025 shocked the global AI industry with frontier-quality results at a fraction of the training cost. The company announced a shift away from pure foundation model scaling toward two narrower bets: a Sea AI initiative targeting Asia-Pacific markets outside mainland China, and commercial application AI for enterprise customers. The consumer Wanzhi productivity app received reduced investment as part of this refocus. This is not a stealth startup. It is a publicly profiled company led by one of the most recognizable Chinese-American technology figures, operating in an extraordinarily competitive domestic market while navigating US export controls on advanced GPUs, reportedly sourcing H100, H800, and H20 inventory through Singapore and other channels before the controls tightened. Hiring continues, but the bar and the strategic context have both shifted meaningfully since the early hyper-growth phase.

Application Process

  1. 1
    Identify the right channel: 01

    Identify the right channel: 01.AI posts roles on its corporate site (career.01.ai or the careers section under 01.ai), on LinkedIn for international and Singapore roles, and on Liepin, BOSS Zhipin, and Maimai for mainland China engineering positions.

  2. 2
    Tailor a single-page Chinese resume for Beijing or Shanghai roles and a clean on

    Tailor a single-page Chinese resume for Beijing or Shanghai roles and a clean one to two page English resume for Singapore, research, and international positions; do not submit a generic Western resume to a Beijing posting.

  3. 3
    Front-load the highest signal item: a published paper at NeurIPS, ICML, ICLR, AC

    Front-load the highest signal item: a published paper at NeurIPS, ICML, ICLR, ACL, EMNLP, or CVPR; an open-source model or library with real adoption; or production experience training, fine-tuning, or serving large language models at scale.

  4. 4
    Apply through the official site or a verified recruiter introduction; cold appli

    Apply through the official site or a verified recruiter introduction; cold applications via Maimai messages to 01.AI engineers can work but are noisier than a referral from someone inside the Yi team.

  5. 5
    Expect a recruiter screen within roughly one to two weeks if your profile matche

    Expect a recruiter screen within roughly one to two weeks if your profile matches; international candidates may move faster due to the smaller Singapore pipeline.

  6. 6
    Prepare for a technical screen of around sixty minutes covering coding (LeetCode

    Prepare for a technical screen of around sixty minutes covering coding (LeetCode medium to hard, frequently with a systems twist), then two to four onsite or virtual onsite rounds that mix coding, ML system design, and a deep dive into your most recent work.

  7. 7
    For research roles, expect a paper presentation round where you walk through a p

    For research roles, expect a paper presentation round where you walk through a publication or project end to end and defend methodology choices against pointed questions.

  8. 8
    A final round with a senior leader, a director, or in some cases Kai-Fu Lee hims

    A final round with a senior leader, a director, or in some cases Kai-Fu Lee himself for senior hires, focuses on motivation, fit with the post-pivot strategy, and willingness to operate in a fast, demanding environment.

  9. 9
    Offer timelines vary; expect roughly two to six weeks from first screen to offer

    Offer timelines vary; expect roughly two to six weeks from first screen to offer, with compensation negotiation conducted directly with HR and the hiring manager.

  10. 10
    Foreign hires for Beijing should plan for visa and work-permit processing of one

    Foreign hires for Beijing should plan for visa and work-permit processing of one to three months, plus relocation logistics; Singapore roles use Employment Pass under the standard MOM process and tend to be faster.


Resume Tips for 01.AI

recommended

Lead with quantified, model-relevant impact: parameter counts trained, tokens pr

Lead with quantified, model-relevant impact: parameter counts trained, tokens processed, throughput improvements, evaluation benchmark deltas, latency reductions, or inference cost reductions, not vague claims about leading a team.

recommended

Name specific frameworks and infrastructure honestly: PyTorch, JAX, Megatron, De

Name specific frameworks and infrastructure honestly: PyTorch, JAX, Megatron, DeepSpeed, vLLM, SGLang, FlashAttention, Triton, CUDA, NCCL; do not pad with tools you have only touched superficially.

recommended

If you have worked on Chinese language modeling, tokenization, evaluation harnes

If you have worked on Chinese language modeling, tokenization, evaluation harnesses, or RLHF for Chinese data, surface this prominently because it is a direct match for 01.AI's bilingual focus.

recommended

Cite open-source contributions with links: a merged PR to Hugging Face Transform

Cite open-source contributions with links: a merged PR to Hugging Face Transformers, vLLM, or a model release on Hugging Face Hub carries more weight than internal-only project descriptions.

recommended

For research candidates, list publications in standard academic format with venu

For research candidates, list publications in standard academic format with venue, year, and a one-line plain-language description of the contribution; include arXiv links for unpublished or preprint work.

recommended

Show GPU-scale experience explicitly when applicable: training runs on hundreds

Show GPU-scale experience explicitly when applicable: training runs on hundreds or thousands of H100, H800, A100, or H20 nodes is a stronger signal than abstract claims about distributed training.

recommended

Submit your resume in both Chinese and English when applying to Beijing or Shang

Submit your resume in both Chinese and English when applying to Beijing or Shanghai roles; even strong English-only candidates can be filtered by Chinese-speaking recruiters who skim for Chinese-language credentials.

recommended

For product, design, business, or operations roles, frame impact in revenue, ret

For product, design, business, or operations roles, frame impact in revenue, retention, enterprise customer wins, or measurable launch outcomes; vague responsibility statements are heavily discounted at a startup with strict bar pressure.

recommended

Avoid AI-generated buzzword resumes; reviewers at 01

Avoid AI-generated buzzword resumes; reviewers at 01.AI screen large numbers of LLM-augmented submissions and have grown skilled at spotting them.

recommended

Keep formatting clean and plain: a single-column PDF in Chinese for domestic rol

Keep formatting clean and plain: a single-column PDF in Chinese for domestic roles or a clean English template for international roles, no graphics, no photos for international applications, and no infographic styling.



Interview Culture

01.AI's interview culture mirrors a fast-moving Beijing AI lab more than a polished Western corporate process.

Loops are technically demanding, sometimes scheduled with short notice, and weighted heavily toward demonstrated craft rather than credentials. For research and infrastructure roles, expect detailed questioning on training stability, optimizer choices, attention variants, parallelism strategy across data, tensor, pipeline, and expert dimensions, evaluation methodology, RLHF and DPO trade-offs, inference optimization, and the specific reasons behind the Yi team's published architectural choices. Product, design, and business interviews emphasize directness. Reviewers ask why you chose a specific approach, what you would do differently, and how you measure success in concrete terms. Hand-waving is a fast disqualifier. Senior rounds with directors or with Kai-Fu Lee himself tend to focus on motivation, alignment with the post-pivot strategy of Sea AI and commercial applications, willingness to operate at startup intensity, and a candid view of where the candidate's strengths and gaps actually lie. Internal communication blends Mandarin and English depending on team and seniority. Kai-Fu Lee's leadership style is more Anglophone-friendly than the average Beijing tech company, so foreign hires can function in English in many meetings, but Mandarin is essential for full integration into mainland engineering teams, customer conversations, and informal channels. Singapore-based roles operate primarily in English with Mandarin as a useful secondary asset. Work intensity is real. Even with Beijing's 2023 Supreme People's Court ruling reinforcing that 996 work schedules are illegal, startup norms in Chinese AI labs still skew toward long hours, especially during model release cycles and customer pushes. Be honest with yourself about whether that pace fits your life right now and ask explicit questions about expected hours, on-call, and release cadence in your loop.

What 01.AI Looks For

  • Demonstrated craft in large-model work: training, fine-tuning, evaluating, or serving LLMs at meaningful scale, with clear evidence of personal contribution rather than committee credit.
  • Strong fundamentals in deep learning, optimization, distributed systems, and CUDA or accelerator-level performance work for engineering and research candidates.
  • Bilingual capability where the role requires it: Mandarin Chinese for Beijing and Shanghai integration, English for Singapore, research, and international collaboration; biliterate candidates are particularly valued.
  • Open-source credibility: real contributions to PyTorch, vLLM, Hugging Face Transformers, SGLang, Megatron, or your own released models with measurable adoption.
  • Publication record at top-tier venues for research roles, including reproducible code and honest reporting of negative results, ablations, and known limitations.
  • Commercial and applied judgment for product, design, and business hires: ability to translate model capability into customer outcomes in either Asia-Pacific consumer or enterprise contexts after the 2025 strategic shift.
  • Resilience and intellectual honesty about post-DeepSeek competitive realities, the foundation-model capital race, and the company's own pivot away from pure scale.
  • Willingness to operate at startup pace with ambiguous priorities, frequent reorganizations as the strategy refines, and an evolving product surface.
  • Cultural alignment with Kai-Fu Lee's stated philosophy: serious about AI's potential, candid about its limits, focused on impact rather than hype.
  • Long-term commitment signals; 01.AI has been burned, like many Chinese AI labs, by talent churn driven by aggressive poaching from BAT incumbents and well-funded peers.

Frequently Asked Questions

Is 01.AI still a good place to build a frontier AI career after the 2025 pivot?
It depends on what you want. The pivot away from pure foundation model scaling means fewer huge pre-training runs from scratch, but real work continues on Yi-Lightning and successor models, on enterprise deployment, and on the Sea AI regional product. If your goal is to spend three years on a single trillion-parameter pre-training effort, larger labs are a better fit. If you want serious applied LLM work in a small, recognizable team, 01.AI remains a credible choice.
Do I need to speak Mandarin to work at 01.AI?
For Beijing and Shanghai engineering, research, product, and business roles, Mandarin is effectively required for full integration even though some technical conversations happen in English. For Singapore-based roles tied to the Sea AI initiative, English is the working language and Mandarin is a useful but not strictly required asset. Be honest with yourself; partial Mandarin is sometimes harder than none.
How does 01.AI compare to Alibaba Qwen, Baidu ERNIE, ByteDance Doubao, and DeepSeek as an employer?
01.AI is smaller and more focused than the BAT incumbents and trades scale for ownership and speed. Compared to DeepSeek, 01.AI has greater public profile and a more international posture under Kai-Fu Lee, while DeepSeek has captured recent technical mindshare. Compensation at the BAT giants is generally higher and more stable; 01.AI offers more equity exposure, more ambiguity, and more direct contact with senior leadership.
What is Kai-Fu Lee actually like as a leader?
Publicly, Kai-Fu Lee is one of the most articulate Chinese-American AI figures, comfortable in both English and Mandarin, with a deep network across Silicon Valley and Beijing. Internally, accounts describe a demanding but Anglophone-friendly culture with high expectations on craft and clear willingness to make hard strategic calls, including the 2025 pivot itself. Treat any specific claim about him from third parties skeptically and ask candid questions during your loop.
What does compensation look like at 01.AI?
Compensation is competitive within the Chinese AI startup market but below US frontier labs in raw cash. Senior engineers and researchers in Beijing can expect total cash packages in the high six to low seven figures RMB depending on level, with equity in a privately held unicorn that carries both genuine upside and meaningful risk after the 2025 strategic narrowing. Singapore packages are denominated in SGD with a different cost structure. Always ask for written details and benchmark against peer offers.
How real are 996 hours at 01.AI?
Beijing AI startups in general still trend toward long hours despite the Supreme People's Court's 2023 ruling that 996 schedules are illegal. At 01.AI specifically, intensity peaks around model releases and customer pushes. Day-to-day rhythm varies by team and manager. Ask in your loop about realistic working hours, weekend expectations, and on-call structure, and weight the answers from line engineers more heavily than from recruiters.
How does 01.AI handle US GPU export controls?
01.AI is not on the US Entity List, but acquiring advanced accelerators under the tightened export-control regime is operationally complex. The company has reportedly used H100, H800, and the China-spec H20 across various phases, with sourcing strategies including Singapore and other Asia-Pacific channels discussed in 2024 industry reporting. Treat publicly reported details as approximations; if you join an infrastructure role you will see the real picture under NDA.
What is the difference between Yi open-source models and Yi-Lightning?
The Yi-34B and Yi-1.5 family are released under the permissive MIT license and are freely available for download, fine-tuning, and commercial use, which built early developer goodwill. Yi-Lightning, released October 2024, is a closed-source production model that briefly ranked sixth on the LMSYS Chatbot Arena and serves as 01.AI's frontier offering, accessed through the company's API rather than weight downloads.
Can foreign candidates without Chinese language skills get hired?
Yes for the Singapore office and for some senior research, infrastructure, or international product roles based or able to operate in English. No realistically for most Beijing or Shanghai engineering and product positions, where day-to-day collaboration, customer conversations, and informal communication happen in Mandarin. Be candid about your language level when applying.
What is the visa and relocation experience like?
For Beijing roles, expect roughly one to three months for the China Z work visa and residence permit process, plus household relocation and apartment search. The company supports relocation for senior hires but the overall friction is real. For Singapore roles, the Employment Pass process is faster and well established under MOM rules. Ask about specific support, relocation budget, and family considerations early in the conversation.
How stable is 01.AI given the 2025 strategic shift?
Funded with several hundred million dollars and led by an experienced founder, 01.AI is not facing immediate runway risk publicly disclosed, but the strategic narrowing reflects real market pressure from DeepSeek, Qwen, Doubao, and ERNIE. Stability at any post-pivot startup is conditional on the new strategy working. Treat job security claims with appropriate skepticism, ask about runway and revenue trajectory under NDA in late-stage interviews, and keep your network warm.
Should I expect to work directly with Kai-Fu Lee?
For senior hires, yes; he is reportedly involved in final-round interviews and strategic decisions. For mid-level engineers and individual contributors, contact is more episodic, typically through all-hands meetings, occasional reviews, and topic-specific conversations rather than daily collaboration. If working directly with the founder is important to you, ask your hiring manager honestly about expected interaction patterns at your level.

Check Your Resume Before Applying → Review 01.AI role context

Related Resources

Similar Companies

Related Articles


Sources

  1. 01.AI official website
  2. Kai-Fu Lee on Wikipedia
  3. 01.AI on Wikipedia
  4. Reuters coverage of 01.AI Series A funding round, November 2023
  5. Yi-34B model card on Hugging Face
  6. LMSYS Chatbot Arena leaderboard
  7. Reuters reporting on 01.AI strategic shift in early 2025
  8. DeepSeek V3 technical report on arXiv
  9. DeepSeek R1 technical report on arXiv
  10. Sinovation Ventures official site
  11. Bureau of Industry and Security advanced computing chip export control rule
  12. China Supreme People's Court 2023 guidance on 996 work hours
  13. Singapore Ministry of Manpower Employment Pass eligibility
  14. Hugging Face Open LLM Leaderboard archive