This is where we'll write down what we're learning while we build MpegFlow.
The format will be specific, not generic. We're a small team building a video infrastructure product, and the parts that are interesting are the parts most posts about video infrastructure leave out: the trade-offs, the failures, the "this should have been obvious but wasn't" moments.
What you'll find here
A few categories of post we're committing to.
Architecture decisions, with the alternatives we rejected
Every architecture choice is more interesting in the context of what you didn't pick. When we explain why pipelines are modeled as a DAG instead of an imperative script, we'll spend the post on the comparison, not the result.
Production lessons from running FFmpeg at scale
FFmpeg is the workhorse, but the tutorials don't cover the operational parts: queue back-pressure, retry semantics, audit-trail design, versioning encoder presets in production. That's the bulk of what we've been building. We'll write about it.
Honest comparisons with adjacent tools
There are good products in the video-infrastructure space — Bitmovin, AWS Elemental, Mux, Cloudflare Stream. We're not going to pretend they don't exist or that we beat them on every axis. The most useful comparison posts are the ones that say what each tool is actually good at.
What you won't find here
- "5 ways AI is revolutionizing video processing"
- "Why MpegFlow is the future of broadcast"
- "Customer success story: how X used MpegFlow to scale 10×"
We'll get to product-marketing content eventually. Not yet, and not here.
How to follow along
- RSS: /blog/feed.xml — proper Atom feed, syndication-friendly
- Beta list: join the cohort and we'll email you when there's something worth reading. Two emails a month, max. Reply STOP and you're out.
- GitHub: the engineering posts will often link to specific commits and PRs in our repo as primary sources. We're trying to keep the blog and the code honest about each other.
That's it for the meta-post. Real content next.