Applied Research Scientist

Location: Team Engineering

Team: Country Beijing / China

About the Role

At Canva, we're democratizing design through magical AI experiences that empower hundreds of millions of users worldwide. As an Applied Research Scientist in our Beijing Engineering team, you'll pioneer the next generation of AI-powered creativity, blending cutting-edge research with Canva's collaborative, design-first culture to build tools that feel truly intuitive and impactful. In this role, you'll own application-focused research, rapidly prototyping and validating the latest generative models like FLUX, Qwen series, and DeepSeek to deliver production-ready innovations. You'll design novel tasks that push multimodal AI boundaries—think agentic systems for DeepResearch, character-consistent visuals, and glyph-perfect typography—while collaborating with top ML engineers and product teams to integrate them seamlessly into Canva's ecosystem. Whether through SFT for aesthetic enhancements or efficiency optimizations like distillation, your work will directly enhance user productivity in design. We're seeking passionate researchers at all levels, from experienced hires to 2025/2026 graduates and interns, who thrive in our onsite Beijing office. With a collaborative mindset, you'll partner across teams, document breakthroughs, and contribute to papers or talks, all while enjoying Canva's innovative perks like unlimited vacation, learning stipends, and a culture that celebrates bold ideas. Join us to revolutionize how the world creates!

Key Responsibilities

• Conduct application-focused research on latest generative AI advancements and validate their impact
• Turn fundamental generative model outcomes into high-impact Canva applications through innovation
• Design new frontier tasks for multimodal AI models to expand capabilities for design productivity
• Collaborate with ML Engineers to identify trends and explore new application scenarios
• Perform SFT and post-training on models to address user needs like character consistency and visual aesthetics
• Deploy and optimize diffusion models and LLMs for production at Canva scale
• Partner closely with engineering and product teams to integrate AI into Canva's ecosystem
• Accelerate model inference efficiency using techniques like distillation and quantization
• Document research findings and share knowledge internally and externally
• Experiment with cutting-edge AI techniques to advance Canva's magical design experiences
• Lead efforts to bring best-in-class AI models to hundreds of millions of Canva users

Required Qualifications

• Rich experience in shipping latest generative models like FLUX (Diffusion Transformer), Qwen3-Coder, Qwen2.5-VL, and DeepSeek-R1
• Proven ability to translate AI research advancements into robust, production-ready implementations
• Hands-on experience deploying diffusion models and LLMs with efficiency improvements via distillation, speculative sampling, or quantization
• Strong proficiency in Python, PyTorch, and cloud computing platforms for fast model deployment
• Collaborative experience working with engineers, product teams, and researchers
• Bachelor's or higher degree in Computer Science, Machine Learning, or related field (open to all experience levels including 2025/2026 graduates and interns)
• Track record of sharing knowledge through documentation, internal collaborations, papers, or talks

Preferred Qualifications

• Experience in agentic AI models and multimodal generation for productivity applications
• Background in designing novel AI tasks aligned with real user needs in design tools
• Familiarity with post-training techniques like SFT for character consistency, glyph generation, and visual aesthetics
• Prior work in creative AI applications for design platforms
• Publications or contributions to open-source AI projects in generative models

Required Skills

• Generative AI models (Diffusion Transformers, LLMs)
• Multimodal AI (vision-language models like Qwen2.5-VL)
• Model deployment and inference optimization
• Supervised Fine-Tuning (SFT) and post-training
• Python programming
• PyTorch framework
• Cloud computing (AWS/GCP/Azure)
• Distillation and quantization techniques
• Speculative sampling strategies
• Agentic AI systems
• Cross-functional collaboration
• Research documentation and communication
• Problem-solving in production environments
• Creative thinking for design applications
• Trend tracking in AI advancements

Benefits

• Work onsite in Beijing with Canva's vibrant China team
• Flexible full-time schedule with opportunities for interns and new grads
• Canva's unlimited vacation policy and recharge days
• Comprehensive health insurance and wellness programs
• Learning stipend for conferences, courses, and certifications
• Stock options and performance bonuses
• Parental leave and family support benefits
• Free snacks, meals, and team-building events
• Collaborative, design-forward culture with global impact

Canva is an equal opportunity employer.