s
s
CLOUD AI RESEARCH
Academic Minimalism
Research notes on cloud-native AI systems, distributed training, and infrastructure.
Writing by year
No thumbnails. Just text.A research note on making distributed AI services observable, recoverable, and easy to reason about.
CloudAIDistributedSystemsReliability
hello
hello
A note on why training jobs, schedulers, and observability should be designed together.
CloudInfraDistributedTrainingObservability
How I annotate OSDI papers to recover the design constraints hidden between the lines.
PaperReviewSystemsResearchWorkflow
A short note on treating observability as a first-class training primitive rather than a postmortem tool.
DistributedTrainingObservabilityMLOps
A lightweight framework for deciding when to retrain versus when to recalibrate.
EdgeAIEvaluationMLOps
A note on why accuracy alone is not enough when inference has to live on constrained devices and unstable networks.
EdgeAIInferenceSystemsDesign