Designs

Company tags are community-reported. Counts on cards show how many people reported that design.

• 

  102. Design a GPU Cluster Scheduler for Training Jobs

  Thousands of GPUs shared by many teams: gang admission, topology-aware placement, quotas with lending, checkpointed preemption and failure recovery.

  ClassicMedium

  AI infrastructureAWSGPUs

  +2

  KubernetesScheduling

  Worked design

  Practice
• 

  103. Design Model Weight Distribution to a GPU Fleet

  A 500 GB model on 1,000 GPU servers in minutes: chunks and hashes, a topology-aware swarm, signed manifests, bandwidth budgets and waves.

  ClassicMedium

  AI infrastructureAWSDistribution

  +2

  InfrastructurePeer-to-peer

  Worked design

  Practice
• 

  117. Design a Self-Healing GPU Fleet

  Detect, classify and repair failing GPU nodes (Xid errors, ECC, NVLink and EFA faults, stragglers, silent corruption) without draining the fleet.

  ClassicMedium

  ObservabilityAI infrastructureAWS

  +2

  GPUsReliability

  Worked design

  Practice

Each design brings together the brief, practice, examples and discussion.