Meridian Signal Lab began in 2015 as a place a few engineers kept re-explaining the same ideas to each other. We cleaned it up and opened the door.
We write educational material on signal processing, computer vision and applied machine learning. The format is the explainer: one idea, walked from intuition to implementation without a gap where you quietly lose the thread. We are deliberately not a bootcamp, a consultancy funnel or a newsletter growth machine — the writing is the product, and it stays free.
Every entry follows the same discipline. State the problem in a sentence. Build the intuition with the smallest example that still holds the idea. Show a minimal, runnable check where one exists. Then name the caveats out loud. If a claim can't be verified with at least a small experiment, we mark it untested rather than assert it. And we revise — the review date on a piece matters as much as the publish date.
We diagram things because most of these ideas are spatial before they are symbolic. A signal splitting into frequencies, a bounding box drifting across frames, a curve bending to chase noise — these are pictures first. The notation is a compression of the picture, and we think it's easier to decompress the notation once you've seen what it's a picture of.
Clear notes make the whole field a little less gatekept. That's the entire business model, and we're comfortable with it. When we mention a paid book or tool, we say so plainly and mark any affiliate link; a commission never changes what we recommend.
We keep one local key to remember your cookie choice. Cookie notice.
