I wouldn't put the time into making your own ICCs as it will not improve your proof observation environment. It won't make any difference. The human brain can not adapt to two different white points and assume matches are met. We just do not function like that on a physiological level. And you can not adapt the 3500k to your display - that is too far out of the ability for ICC conversion to occur accurately. So you are in effect, comparing apples to oranges right now - although it sounds that the contrast level of the display is working.
The goal in color management is to keep a consistent white point. If you simultaneously look at white light that is neutral (5000k) and look at white light that is cool (maybe 6500k) and look at white light that is warm (maybe 3500k), your brain does not see three white lights. If you only look at warm white light, your brain will see it as white. If you only look at cool white light, your brain will see it is as white. But, as soon as you turn on two different color white lights - your brain sees two different colors and the same is true when comparing colors and grays under two different colors of white light. This is how the standard observer operates because of the beauty of or the limitations of our visual system.
In our studio at Cone Editions Press, we have 5000k everywhere you set something down. The entire studios on two floors are 5000k. The viewing booths are 5000k and our imaging rooms are dark with the exception of the illumination from the displays and the dimming viewing booths. But, on our main studio wall we have warm or neutral lights. The warm light is to imitate what happens in gallery settings. Ultimately when we are working with a photographer or artist we try to simulate the situation in which the prints will ultimately be viewed. If very bright, we may image differently. If very dim (as in a museum) we would most definitely image differently. There have been times when we have made the viewing wall the same color as the gallery and used the exact same lighting. For that particular project, it mattered that much to the photographer and the work was very much nuance based.
So you either have to wing it or think about finishing off your imaging area with a dimming GTI viewing booth. Then you can get an uncanny match. But that comes at a cost and you need to decide whether its worth it to you. Sometimes, the best way is the old way. You just go out into the wall and look at the proof and mark it up and go into Photoshop and make the requisite changes. That way you spend more time looking at the proofs and less time looking at Photoshop. We're quickly becoming however, a display generation. So I vote for marking up your proofs!