Multimedia Engineer (Mandarin Speaking) - Sydney

About the role

TikTok's Multimedia Platform department builds multimedia processing systems such as video transcoding, cloud editing and image processing platforms, handling hundreds of millions of videos and tens of billions of images daily. It is looking for an experienced Multimedia Engineer to join its Sydney team, working on large-scale multimedia processing systems and high-performance computing frameworks, collaborating with cross-functional teams and solving engineering problems across multimedia, heterogeneous computing and AI infrastructure.

Key responsibilities

• Take part in developing the video computing framework to support efficient processing of the company's massive video and complex business needs.
• Support product-line requirements such as image processing, video transcoding, video editing and video analysis.
• Handle performance analysis and optimization of video processing.
• Help design and optimize large-scale multimedia processing systems to improve throughput, latency, stability and resource utilization.
• Work with product, algorithm, platform and infrastructure teams on multimedia and AI engineering solutions.

Required profile

• Bachelor's degree or above in Software Development, Computer Science, Computer Engineering, Electrical Engineering or a related field.
• 2+ years of hands-on experience in software engineering, multimedia systems, high-performance computing, backend infrastructure, heterogeneous computing or related areas.
• Proficiency in Chinese and English, as Mandarin is a primary working language for global partners in Mandarin-speaking markets.

Required skills

• Proficiency in C/C++, Python and/or Go, with strong coding, debugging and problem-solving skills.
• Solid understanding of operating systems, data structures, algorithms, networking and concurrent programming.
• Preferred: experience with video codecs, media frameworks, performance optimization, transcoding platforms, GPU hardware and inference stacks such as CUDA or CUTLASS.
• Preferred: knowledge of LLM and multimodal models, inference frameworks such as vLLM and SGLang, and model quantization.