Ilan Lewin
Ilan LewinPhysicist, engineer, three-time founder. Amstelveen.

Getting your measurement code off MATLAB and LabVIEW

If I sent you this, it's probably because something important in your company runs on a MATLAB script or a LabVIEW rig, and only one or two people can really change it.

I've been on both sides of that. I've written the lab code that proved a measurement worked, and I've been the one who had to turn it into a product other engineers could own. Thirty years of it: FPGA and RF at Netline, a multi-sensor platform at Kineti, robots now.

What usually hurts

How I'd get you out

  1. Find the part that matters. Usually a small slice of the code carries the actual physics.
  2. Record what the current system does on your real data. That becomes the test the new code has to pass, number for number.
  3. Port it piece by piece to Python and C. AI does most of the reading and translating, the tests decide what's right.
  4. Run old and new side by side until they agree. Your instruments mostly stay, they work fine with standard drivers.

The part I'd actually sell you on

I do the work with an AI setup built around your code, and when I leave, it stays. Your engineers can ask it why a coefficient is 0.87, trace a bug from a bad capture, or try a change and have the old system's recorded results check it.

It lives in your repo, runs on a model you pick (on your own servers if the code can't leave the building), and there's no per-seat fee. The person who knew everything gets their afternoons back.

How we'd start

Half a day with whoever knows the system, no charge. After that I can tell you what it would take, or that you're better off leaving it alone. If we go ahead, I work inside your team one or two days a week.

Send me a line on what the system does and roughly how old it is. That's usually enough to know if it's worth a call.

ilan@aardvark-sensing.com  ·  LinkedIn

Ilan