AI大模型算法｜LLM/VLM Pre-training

Canva is hiring an AI大模型算法｜LLM/VLM Pre-training in Beijing, China. Level rates it ; you can apply on Canva.

Level's read on this job

On Level's scale, this role is Builds AI ●●●●: the job is building AI systems.
Tools and skills the posting cites include Deepseek, PyTorch and Deep Speed.
Level classifies it as a senior role.
Level lists 77 other open roles at Canva.
The role was posted in the last few days.

AI in this role

Develop frontier language and multimodal generative models using high-performance GPU clusters at Canva's Beijing AI team.

deepseekpytorchdeep-speedmegatron

llmvlmpre-trainingreinforcement-learningmultimodal

Join Canva’s Beijing AI team to develop frontier language and multimodal generative models using high-performance GPU clusters and high-quality datasets. You’ll collaborate with research, engineering, and product teams to bring research advances into products that empower hundreds of millions of users.

What you’ll do

Define and lead research initiatives in LLM / VLM pre-training and multimodal generation, informed by frontier research and product needs.
Design and run large-scale training experiments, analyze results, and improve model capabilities, training efficiency, and stability.
Explore model architectures, training methods, post-training and reinforcement learning algorithms, and high-quality synthetic data pipelines.
Collaborate across teams to bring models into production and share findings through technical documentation, publications, and presentations.

Core pre-training expertise gained on teams such as Qwen, Kimi, GLM, or DeepSeek.
Experience with autoregressive or diffusion models for text-to-image, video, or 3D generation.
Deep expertise in multimodal architectures, pre-training data, reinforcement learning, or synthetic data.

How we rate this

AI大模型算法｜LLM/VLM Pre-training at Canva rates 100 out of 100 for how much of the daily work is AI. That makes it Builds AI (AI Level 4 of 4). The level is about AI in the job, not seniority.

Classification

Builds AI. The job is building AI systems.

●●●● Builds AI80 to 100
●●●○ Works on AI60 to 79
●●○○ Uses AI40 to 59
●○○○ Little AI0 to 39

Levels come from how often the tools, models and workflows of the role are named in the posting itself. Read the description and count the AI tools, models and workflows it names.

Key facts

AI大模型算法｜LLM/VLM Pre-training is an open job at Canva.
Level rates this role AI Level 4 of 4: Builds AI.
The role is based in Beijing, Beijing, China.
The job was posted on October 28, 2026.
Seniority: senior.
Applications are handled by Canva. Apply on Canva

Prepare for this job

Built from this posting and from similar jobs on Level.

Skills and AI tools this role asks for

LLMVlmPre TrainingReinforcement learningMultimodalDeepseekPyTorchDeep Speed

Questions you could be asked

Tell me about a project where llm was part of your work. What did you do?
Tell me about a project where vlm was part of your work. What did you do?
Tell me about a project where pre training was part of your work. What did you do?
Tell me about a project where reinforcement learning was part of your work. What did you do?
Tell me about a project where multimodal was part of your work. What did you do?

Adapt your resume

List these exact terms on your resume: LLM, Vlm, Pre Training, Reinforcement learning, and Multimodal. An applicant tracking system matches the wording, not the idea.
Attach one line of real, concrete experience to at least one of them. A tool named with nothing behind it rarely survives a human read.
Lead with what you built, trained or shipped. This role is judged on the AI system itself, not the tools around it.

Want an expert to read your CV for this job?

Free. Send your CV and the role you want next. We reply by email within 2 to 4 business days.

Get a free CV review Get your CV rewritten Check your resume with the ATS resume checker