▸ Blog · Page 6
More from the trenches.
Older posts. Same hot takes.
-
DevOpsDitch expensive cloud providers for self-hosted SaaS
The cloud tax kills SaaS margins early. How self-hosted Coolify on cheap ARM instances at Hetzner or Oracle slashes a $500 AWS bill toward zero.
Read post →
-
LaravelLaravel Octane for high-traffic PHP apps
Laravel Octane boots your app once and keeps it warm in memory, feeding requests through a worker pool. How to cut boot overhead and scale to high traffic.
Read post →
-
ShopifyShopify UCP quick-start: make your store agent-ready
AI agents are the new buyers, and they can't see your store without a UCP manifest. A quick-start to making your Shopify store agent-ready.
Read post →
-
AIClaude MCP: connecting my dev tools to LLMs
The Model Context Protocol is a USB-C port for LLMs: one MCP server, any host, no MxN integration tax. The architecture, the servers I run, and why it is safe.
Read post →
-
ArchitectureCaching for speed: Redis and semantic layers in RAG
Stop paying for the same LLM call twice. Two-tier caching with Redis keys and RedisVL semantic lookups slashes RAG latency and trims your LLM API bill.
Read post →
-
ArchitectureSmart auto-scaling for modern AI apps
CPU auto-scaling fails GPU workloads. Why queue depth, KV-cache pressure, and TTFT beat CPU as triggers, plus the KEDA patterns to scale AI in time.
Read post →
-
ArchitectureGPU-aware load balancing for AI inference
Round-robin breaks when LLM requests span 50 to 50,000 tokens. GPU-aware load balancing with prefill/decode disaggregation and the metrics that cut P99.
Read post →
-
ArchitectureCircuit breakers: stopping vector DB failures
A slow vector DB kills SaaS faster than a dead one. The circuit-breaker pattern for AI infra — states, fallback tiers, and Laravel-friendly wiring.
Read post →
-
ArchitectureMessage queues for heavy-duty document processing
Stop running embeddings in the request cycle. Build a document pipeline on message queues with staged workers, retries, dead-letter queues, and autoscaling.
Read post →