Описание: FAR Labs is a unified platform for high-performance AI inference. It provides dedicated Inference and Model APIs with flexible workload placement across its infrastructure, a customer's cloud or hardware, and shared GPUs.
Задачи:
•Set the technical direction for the inference stack and benchmarking practices
•Improve the speed and efficiency of model serving
•Validate performance gains with credible benchmarks
•Lead the engineers building the platform
•Shape the technology, performance standards, and engineering culture behind FAR Labs
Требования:
•Production experience with LLM inference-serving systems
•Expertise with vLLM, SGLang, TensorRT-LLM, or similar stacks
•Strong knowledge of batching, paged attention, KV-cache management, and modern serving architectures
•GPU performance engineering experience with CUDA or Triton
•Solid understanding of GPU utilisation, memory bandwidth, and precisions such as FP8 and FP4
•English B2
Условия: No conditions specified
