There should be info on ovens on-line somewhere, and again the level of crazy is up to you. My old JRL standard cell oven consisted of an inner box, aluminum, with some resistance wire wrapped around it, or maybe flat heater pads. Control was with a mercury thermoregulator, but today you'd use a thermistor or other sensor of your choice, and a control circuit. They then put that inside another aluminum box, wrapped with resistance wire and controlled the same way. The inner oven ran a bit hot, 37C I think, and the outer oven a few degrees less. It's like using double voltage regulators- no ambient is going to affect that inner box. I've built simple single ovens that held a few millidegrees, and I don't think that's needed here. Look for a Jim Williams Linear Technology app note with a temp control circuit. Hewlett Packard made a 1N821 based voltage standard years ago, visually similar to the Fluke 731- so old I can't find a picture of one. The diode circuit was on a small and not-too-special pcb that lived inside a small drawn aluminum can- wrapped with resistance wire and stuffed in a block of foam for insulation. You just need to get a few degrees higher than the highest ambient you ever expect, and hold it within a tenth degree or so. If I were doing one today I'd probably get some 1" copper pipe and copper end caps. Wrap it with insulated resistance wire (or use a Kapton heater pad- eBay is loaded with them), glue a sensor to it, and make a simple proportion control. I wouldn't use a nice PWM control because of the noise. Install some feedthrus for wire and stick it a block of Styrofoam for insulation.