First articles dropping soon. Subscribe to get notified →

Engineering Notes
from BHK Cloud.

Field reports on GPU hosting, object storage economics, and AI-native platform engineering: what we ship, what we break, and what we learn building infrastructure for serious AI teams.

Browse Articles → Subscribe via email

Topics we cover

GPU Hosting Object Storage AI Economics MLOps Infrastructure Performance Compliance Field Notes

Engineering Notes

Coming soon.

Six articles in draft. Subscribe below to be first in line.

AI-Native Storage·9 min read

Beyond backup: why traditional cloud storage fails the AI-native era

Comparing $0.99/TB zero-egress S3 against Wasabi, Backblaze, and AWS, and what it means for AI economics.

Coming soon
Architecture·11 min read

Designing zero-egress data planes for sovereign AI workloads

How short-lived signed policies let GPU clusters access data in place without crossing region boundaries.

Coming soon
Platform·8 min read

Composable primitives: storage, networking, compute as policy

Treating infrastructure as declarative intent and what we learned wiring 11 regions to a single control plane.

Coming soon
Performance·7 min read

Direct-from-S3 streaming for keeping GPUs saturated

Eliminating redundant caching by streaming shards straight into PyTorch dataloaders.

Coming soon
Field Notes·5 min read

What 30B-parameter customers actually ask about

Cost per token, deterministic run times, and the pricing transparency questions we keep hearing.

Coming soon

Want these in your inbox?

A short engineering note every other week. No marketing fluff, just shipped work and lessons learned.

Subscribe via email → View Pricing