Practical articles on AI, DevOps, Cloud, Linux, and infrastructure engineering.
Fine-tuning is rarely the right answer. We've fine-tuned three times in two years; few-shot or RAG was correct for everything else. The decision criteria.
Standard APM doesn't tell you when your LLM-powered features are silently degrading. The signals we track and the dashboards that catch the regressions standard tools miss.
Evolve CI/CD toward autonomous pipelines that detect issues and roll back safely.
Multi-agent systems are mostly hype. The patterns we've seen actually deliver value, plus the ones we'd avoid until the tooling is more mature.
We have ~40 prompts in production. The patterns that improved quality, the ones that turned out to be folklore, and how we test prompts now.
How we deploy LLM-powered features. The deployment patterns are mostly normal; the validation is where the differences are.
We tried four quantization techniques on Llama-3 and Mistral models. The quality vs cost trade-offs we found, plus what works for production inference.
We benchmarked four vector databases on the same workload. Each has a place. Here's how we'd pick today.
We've shipped four production RAG applications. Each one taught us something. The end-to-end pattern that works.
Run retrieval-augmented generation at scale. Chunking, caching, and observability.
We cut LLM inference cost 47% over a quarter while improving p95 latency. Six changes, ranked by what each one actually delivered.
Wikis rot. We moved every operational doc into the repo it describes. Six months in, the docs are mostly correct because the only people who can update them are the ones who change the system.
