Research Engineer - AI Workload & Systems

About the role

Huawei Canada’s Computing Data Application Acceleration Lab is seeking a Research Engineer to drive AI workload and system research on a 12‑month contract.

Key responsibilities

• Monitor and analyze emerging AI model architectures (LLMs, VLMs, MoE, attention mechanisms) and their computational characteristics for training and inference workloads.
• Develop systematic frameworks to map AI applications to hardware requirements, design efficient inference deployment strategies, and build performance models.
• Evaluate algorithmic and hardware innovations, delivering quantitative architectural recommendations for next‑generation AI accelerators.

Required profile

• Strong understanding of modern AI model architectures and emerging trends such as sparse/linear attention, Mixture‑of‑Experts, and related techniques.
• Solid knowledge of AI accelerator architectures (GPUs, TPUs) with expertise in memory hierarchy, interconnects, and performance bottlenecks.
• Hands‑on experience developing AI kernels using Triton, TileLang, CUDA or equivalent, and familiarity with FlashAttention and LLM inference frameworks (vLLM, SGLang).
• Experience with deep learning frameworks (PyTorch, JAX) and hardware performance analysis tools, including Roofline‑based methodologies.

Required skills

• Triton
• TileLang
• CUDA
• FlashAttention
• vLLM
• SGLang
• PyTorch
• JAX
• Roofline performance analysis
• GPU and TPU accelerator architectures