It sure is possible.

A typical “obscenely bright” LED chip might be Cree XML, but many similar chips exist. You’d need a plano-convex or equivalent Fresnel lens - shorter focal lengths favour compact design. Then you need a driver. Some are fixed while some adjustable with a tiny potentiometer. You’d need an 18650 cell holder (it can be made too, an 18650 will go into a leftover piece of 20 mm electrical cabling pipe with a spring-loaded metal cap engineered of something).

Myself, I bought a nice head lamp, but it broke after one year. The driver board failed. Being of the lazy variety, I replaced the board with a resistor to limit current and now it’s been working 3 years already. Not at peak luminosity, the resistor wasn’t optimal of course. :)

That is quite a lot of interesting experiments, thanks for introducing. :)

I’m inclined to add one more:

51: monitor the radio spectrum for drones (and if their signature looks hostile, warn people about them) - there’s a DIY recipe for a monitoring station out there somewhere, and some Ukrainian guys scan their sky using HackRF

SDR is definitely a technology worth learning. I’m already a happy user of RTL-SDR, but if I want to really see what my WiFi is doing, I should get a HackRF eventually too. (Note: WiFi is too fast to intercept without loss, except with another WiFi card, unless a slower bitrate is deliberately chosen.)

Trying to figure out how my heat pump supposedly supports WiFi… in unfathomable and non-standard ways. It’s available as an access point, I can associate and ping it, but no TCP ports listen and no UDP port responds. Nothing cool, undocumented features down to the rocky bottom. When you buy a heat pump and plan to automate its use, check out supported protocols before making a decision. :)