Miguel is an engineer at camelAI. He writes about data infrastructure, databases, and technical deep dives — including the engineering decisions behind how camelAI is built.
POSTS BY Miguel Salinas · 5
Engineering · Oct 1, 2026 · 6 min read
The web wasn't built for agents. Neither is your app.
We removed the agent loop from camelCode and deleted about 72,000 lines of code. Here's why we built camelRun, an open-source hosted runtime for stateful, long-lived agents.
Engineering · Jul 28, 2026 · 8 min read
Our coding agent runs in a Cloudflare Durable Object, not a VM
We moved camelAI’s coding agent off always-on VMs and into Cloudflare Durable Objects, with SQLite and R2 for files and JavaScript Code Mode instead of bash. Here’s the three-step redesign that made it possible.
Engineering · Jul 22, 2026 · 10 min read
How we self-host DeepSeek V4 Flash on AWS spot instances
We recently launched a free tier powered by a self-hosted DeepSeek V4 Flash. Here's how we picked the model and hardware together, adapted an open source vLLM recipe, and made the KV cache — not raw compute — the thing that let the economics work.
Tips · Jul 23, 2025 · 3 min read
How to Build Chat With Your Data Step by Step: A Practical Guide
AI-powered analytics is transforming user expectations for data interaction. "Chat with your data" capabilities are becoming essential for SaaS platforms, moving beyond static dashboards to conversational interfaces....
Product · Jul 10, 2025 · 2 min read
Which Databases Does Hacker News Hype Most? 1.8M Headlines Exposed
We analyzed 1.8 million Hacker News headlines spanning February 2007 to June 2025 to examine database engine popularity trends. Using camelAI connected to a ClickHouse database, we uncovered which databases the HN...
