Founding Distributed ML Infra Architect

Axiōma Search is seeking a distributed training infrastructure engineer to own end-to-end GPU cluster operations in London, scaling from tens to thousands of GPUs and improving efficiency across the training stack.

You will work hands-on with PyTorch, NCCL, DeepSpeed and Megatron-LM/FSDP on cloud workloads (AWS) and contribute to long-term infrastructure decisions. Visa sponsorship and relocation support available.