Contribute to groundbreaking Physical AI innovations at Veeda AI as a Technical Staff Member. Engage in multi-node video world model training while tackling key performance challenges.
Veeda AI is seeking a Technical Staff Member to enhance its world-class team working with multimodal foundation models. This role requires deep expertise in PyTorch and parallelism stacks like FSDP2 or Megatron-Core. You will optimize distributed training, inference throughput, and ensure numerical stability for cutting-edge AI applications.
Key Responsibilities:



- Own model FLOPs utilization during video world model training - Profile latency and GPU utilization for improved throughput - Enhance precision and stability for video tokenizer and VAE layers - Write and tune custom CUDA and Triton kernels - Develop detection layer for silent data corruption issues
Requirements: - Bachelor's degree or equivalent in a technical field - Hands-on experience with PyTorch in multi-node jobs - Fluency in Python and C++/CUDA programming - Proficiency in profiling live training and inference runs - Expertise in low-precision numerics or kernel authoring
Drive innovation in Physical AI through expertise in PyTorch, distributed systems, and cutting-edge optimizations.#J-18808-Ljbffr

📌 Technical Staff Member For Physical Ai (Winnipeg)
🏢 Veeda AI
📍 Winnipeg