Engineering blogs of data and AI companies

The site tracks 40 of them, from Netflix TechBlog, Airbnb Engineering and Engineering at Meta to Databricks, Snowflake Builders, dbt Labs, ClickHouse and DuckDB on the data side, and OpenAI, Anthropic Engineering and Research, Google DeepMind, Hugging Face and NVIDIA on the AI side. The blogs page ranks them by how often they post and what they write about.

Every blog page lists its feed. Netflix TechBlog is netflixtechblog.com/feed and OpenAI News is openai.com/news/rss.xml. Anthropic Engineering and Anthropic Research publish no feed, so the site reads their listing pages and republishes the titles and links in its own feed at blogs.fru.dev/rss.xml.

Yes. The blogs page has a Download OPML button (blogs.fru.dev/blogs.opml) that Feedly, Inoreader, NetNewsWire, Reeder and most readers import in one step. Or follow blogs.fru.dev/rss.xml for every post in one feed.

Every day at 12:15 UTC the site reads each blog’s feed and adds the new posts. Cadence (posts a month over the last six months) and status (active, quiet or dormant) are worked out from the posts themselves.

By fixed keyword rules over the title, the feed’s own categories and its short summary: AI and agents, machine learning, data engineering, databases, infrastructure, reliability, security, developer tools, research, and product news. No model reads the posts. A post can carry up to three topics.

No. The site keeps the title, the date, the link and at most 240 characters of the summary the blog itself puts in its feed. Every row links to the full post on the company’s own site.

Every blog page has Suggest a correction, and the blogs page has Suggest a missing blog. Each suggestion is checked against the blog before anything changes.