About the Role/Specialty
As an MLE on the Video Effects stream, you’ll design, fine-tune, and deploy models that power Canva’s next-generation video effects, AI editing and interactive experiences. From integrating on-device AI capabilities to building scalable backend pipelines, you’ll bridge research and production — transforming ML ideas into delightful, real-time features for our users.
关于岗位
作为视频特效方向的算法工程师，你将负责设计、微调并部署支持 Canva 下一代视频特效、智能编辑和互动体验的模型。从集成端侧 AI 能力，到构建可扩展的后端流水线，你将连接研究与产品之间的桥梁，把机器学习的创意转化为令人惊喜的实时功能，真正交付给用户。
What you’ll do (responsibilities)
- Design, fine-tune, and deploy ML models for AI-powered visual effects across image and video modalities.
- Build robust pipelines for LoRA-based model training, post-training quantization, and inference optimisation.
- Develop and integrate on-device ML capabilities (MediaPipe, CoreML, NNAPI, etc.) for mobile runtime efficiency.
- Scale backend services for effect model serving, asset management, and real-time rendering.
- Prototype novel AI effects using diffusion, computer vision, and shader-driven approaches.
- Work closely with design and engineering to bring creative ideas to life as user-facing features.
- Stay informed on emerging trends in generative ML, real-time graphics, and VFX workflows to drive innovation.
工作职责
- 设计、微调并部署用于图像与视频 AI 特效的机器学习模型
- 构建稳健的训练流水线，支持 LoRA 微调、量化后处理及推理优化
- 开发并集成端侧 ML 推理能力（如MediaPipe、 CoreML、NNAPI 等），提升移动设备运行效率
- 搭建可扩展的后端服务，用于模型部署、特效资产管理与实时渲染
- 利用 Diffusion 模型、计算机视觉与 Shader 技术，原型化并验证创新 AI 特效
- 与设计与工程团队紧密合作，将创意想法转化为用户可交互的产品功能
- 跟踪生成式 ML、实时图形与视觉特效领域的行业动态，推动产品创新
What we're looking for
We’re looking for an MLE with both technical depth and creative instinct — someone who thrives in fast-moving, cross-disciplinary environments. You’ll bring:
- Academic or practical background in computer graphics, computer vision, or computer animation — with experience in one or more of: numerical methods, geometric processing, physical simulation, character animation, generative models (GAN, VAE), or mobile image processing.
- Proven experience fine-tuning and deploying computer vision models, particularly for real-time or streaming use cases.
- Hands-on skills with LoRA training, model post-processing (quantization, pruning), and on-device inference optimisation.
- Familiarity with image and video ML models and how they integrate into visual or rendering pipelines.
- Proficiency in backend development (e.g., Python, Go) and model-serving architectures.
- Practical knowledge of computer graphics APIs (OpenGL/OpenGLES) and shader development (GLSL/HLSL).
- Experience with AI effect tools, video post-production, or multimedia ML use cases is a strong bonus.
- A mindset that balances research curiosity with rapid product iteration.
- Full proficiency in English communications.
任职要求
- 拥有计算机图形学、计算机视觉或计算机动画的学术或项目经验，熟悉以下一项或多项：数值方法、几何处理、物理仿真、角色动画、生成模型（GAN、VAE）或移动端图像处理
- 有实际微调并部署视觉模型的经验，尤其是面向实时或流媒体场景
- 熟悉 LoRA 训练、模型量化与剪枝，以及端侧推理优化方法
- 理解图像/视频模型与视觉渲染管线的集成方式
- 具备后端开发能力（如 Python 或 Go）及模型服务架构经验
- 熟悉计算机图形 API（如 OpenGL / OpenGLES）及 Shader 开发（GLSL / HLSL）
- 有 AI 特效工具、视频后期制作或多媒体 ML 应用经验者优先
- 兼具研究探索精神与快速产品落地的思维方式
- 具备良好的英文沟通能力
What you will learn and develop at Canva
- Learn to ship creative ML features at scale, powering real-time video editing for millions
- Contribute to an interdisciplinary team bridging product, AI/ML, and design
- Influence Canva’s future in intelligent creative tools through your research and engineering
在这个岗位上，你将获得：
- 掌握大规模发布创意型 ML 功能的能力，助力数亿用户实现实时视频编辑
- 与产品经历、AI科学家和设计师密切协作，参与全流程产品创新
- 让你的研究与工作真正影响 Canva 智能创意工具的未来发展路径
