Microsoft AI
Member of Technical Staff
Frontier AI Infra
- Improve Ray, K8s based training infra stack
- Uncovered performance and init time bottleneck by scale testing workload on CPUs
- Create AI toolchain... more to run infra ablation test to measure performance and accuracy
regression of major infra upgrades
- Measure model MTTF with fault-injection on CPUs
- Improve Ray, K8s based training infra stack
- Uncovered performance and init time bottleneck by scale testing workload on CPUs
- Create AI toolchain... more to run infra ablation test to measure performance and accuracy
regression of major infra upgrades
- Measure model MTTF with fault-injection on CPUs


