Shielded cables may minimise electromagnetic interference (radiated) but not conducted electrical interference - eg, through the 12V supply.
Have you tried small or (and) large caps at the amps/source?
Ditto at the light's power source? (Input & output if it's power module.)
An FYI ramble follows:
Not that I've had to deal with modern noise makers, but things that cause noise can radiate it or send it out along their input (and output) power connections and other wires.
Then it's Catch-22 - the noise-source's wires or circuit radiate, the receiver's (amp/audio source) power and other wires get induced noise or the noise goes through their cases.
Caps can quench the conducted (wire) noise.
Shielding quenches radiated noise.
Shielding means grounding metal cases and maybe adding more metal - maybe wire mesh over ventilation and other "non-metal" holes (or behind plastic displays etc) - basically creating a Faraday cage.
Caps were traditionally small and fast types (0.01uF - 0.1uF ceramic or greencaps), but as I found out recently, sometimes big caps help - eg, 100uF and higher electrolytics.
Sometimes coaxial caps are needed - the cap case is soldered to the equipment case and wires "pass thru" the cap (break wire at the case and solder each end to the respective side - but that's in extreme circumstances like some (IMO poorly designed) 12V PSUs for PCs etc.
And where cases are else need to be floating, small caps from case to GND (or even +V as well) if they can't be hard-grounded - but surely all 12V system cases can be?
Sometimes a cap anywhere between the noise and the receiver is sufficient unless there is bad radiation. In that case, cap as close as possible to the source to quash leakage. Otherwise cap as close to the receiver to quash ingress. Sometimes both.
And sometimes a large and small cap. (Although a large cap should quash the high frequencies done by small caps, due to the materials and construction used, that's usually not the case. EG, big filter caps (10,000uF) at PSU DC outputs with small caps (0.01uF) as well, though the small caps are usually placed near ICs/chips etc so they also quash chip generated noise before radiating from PCB tracks.
There are so many possible sources and solutions. It is often a case of trying this or that and then making it permanent to eliminate the worst (ie, placing caps of different sizes between conductors and GND before soldering or permanently attaching).
Just don't get fixated into one solution or frustrated that
that doesn't work.
Oh - I forgot to mention ground loops etc - typically overcome by "single point grounding". Alas I'm rusty and I hated that crap anyhow...
And I didn't mention series chokes - the equivalent to (large)
parallel capacitors.
But why do people often connect radio and audio direct to the battery? Because the battery is a HUGE cap and is the cleanest DC available (exempting regulation or more caps elsewhere).
Same for noise generators - straight to the battery to squash the noise. (But then noise and receiver conductors are close together, and here we go again - Catch 22.)
Others are likely to have newer and specific solutions, or even correct if I'm wrong above.
[ I remember tin cans over ignition coils and distributor caps soldered to shielding over the HT leads to splugs etc. Even steel wool under hubcaps (ie, bearing hubcaps; not wheel caps). LOL - we still have splus etc yet no longer need such measures... Then came GSM interference... And then Arabic numbering...)
Maybe my best solution - use gas lighting.