AI Platform Architect
GraphcoreAI & company
Austin, United StatesLead
Baillie Gifford
SoftBank Group
Ontario Teachers' Pension Plan
Fidelity International
Schroders
Draper Esprit
Data & AI
About the role
TL;DR
Design and oversee a holistic AI platform for large-scale LLM training and inference.
- •Design and oversee a holistic AI rack-scale platform for trillion-parameter LLM training and high-throughput inference across hardware, software, compute, network, and storage.
- •Key Responsibilities Define end-to-end architecture for highly clustered AI environments to eliminate data bottlenecks.
- •Influence workload orchestration and scheduling using Kubernetes or Slurm for distributed training and inference.
- •Profile and optimise full stack from DL frameworks to OS-level tuning and I/O scheduling.
- •Collaborate with firmware, OS, and hardware teams to drive silicon-influencing strategy and requirements.
- •Requirements Lead/principal-level experience in systems engineering, cloud architecture, HPC, or hardware for large-scale AI platforms (4+ years).
- •Deep knowledge of distributed training (data/tensor/pipeline parallelism) and LLM infrastructure needs.
- •Expertise with system interconnects (PCIe Gen5/6, NVMe, RDMA) and GPU-heavy bare-metal environments.
- •Experience with container orchestration and infrastructure-as-code for GPU/bare-metal deployments.
Required skills
Kubernetes
Domain expertise
ai
Tech stack
Kubernetes