Performance Engineer, GPU

Anthropic is seeking a GPU Performance Engineer to architect and implement foundational systems that power Claude, maximizing GPU utilization and performance at scale. The role involves developing cutting-edge optimizations for large language models, from custom kernel development to distributed system architectures.

• Co-design attention mechanisms and algorithms for next-generation hardware architectures
• Develop custom kernels for emerging quantization formats and mixed-precision techniques
• Design distributed communication strategies for multi-node GPU clusters
• Optimize end-to-end training and inference pipelines for frontier language models
• Build performance modeling frameworks to predict and optimize GPU utilization
• Implement kernel fusion strategies to minimize memory bandwidth bottlenecks
• Create resilient systems for planet-scale distributed training infrastructure
• Profile and eliminate performance bottlenecks in production serving infrastructure
• Partner with hardware vendors to influence future accelerator capabilities

Deep experience with GPU programming and optimization at scale. Strong candidates may have experience with CUDA, Triton, CUTLASS, Flash Attention, tensor core optimization, ML compilers (PyTorch/JAX internals, torch.compile, XLA), performance engineering (kernel fusion, memory bandwidth optimization, Nsight profiling), distributed systems (NCCL, NVLink, collective communication, model parallelism), low-precision techniques (INT8/FP8 quantization), and production systems (large-scale training infrastructure, fault tolerance, cluster orchestration).

Annual salary range: $280,000—$850,000 USD.

Competitive compensation and benefits, optional equity donation matching, generous vacation and parental leave, flexible working hours, and a lovely office space.

Deadline to apply: None. Applications will be reviewed on a rolling basis. Apply through the Greenhouse posting. Visa sponsorship is available.