Developing comments between @AlexP and me into an answer...
As other answers have said, candling of reptilian eggs is possible. For in ovo testing of chicken sex, if the male and female have different pigmentation or other easily-distinguishable features then this can be highly reliable. There is no reason to think this would be different for reptiles.
However, there is an important frame challenge required to this question. The question assumes that genetic randomisation is the dominant feature for reptilian development, and that parents would want to screen their developing eggs for positive/negative features. This is largely true for mammals and birds, because the parent's body keeps the developing embryo at a constant temperature.
Reptiles develop differently though, because the majority of reptiles produce eggs which in nature are left to develop on their own. (Exceptions do exist such as the midwife toad, some frogs or some snakes, of course, but the OP's question assumes a species which does not birth live young.) Since the ambient temperature will naturally vary, reptile DNA contains a large amount of "countermeasures" to vary the embryo's developmental pathways and gene expression depending on temperature. The epigenetic effect of ambient temperature is therefore highly significant, to the extent that the embryo's sex can be dependent primarily on the temperature at which the egg is kept.
For an intelligent species such as the yuan-ti, this has important consequences. It is entirely possible that the physical characteristics of an individual can be predetermined solely by managing the temperature of the egg during development. Humans naturally have not put a great deal of effort into investigating how far this goes, but is likely that intelligent reptiles would have a very long history of discovering what temperature changes at what times during the embryo's development produce what effects - gender as a start, of course, but also physical size and strength, quality of eyesight or hearing, intelligence, or many other possibilities. Candling could be used to screen for developmental abnormalities (embryos which do not look "snaky" enough), but this would be a backup to a much more systematic selection of characteristics for your offspring.
As with pre-medicine human midwifery, of course it is quite likely that many of these would be some kind of tribal knowledge. Some may work, some may only work partially, and some may not work at all. But there would still be a body of knowledge which would be called on, and there would likely be individuals who would fulfil a similar advisory role to human midwives, except on a more continuing basis because eggs need longer-term monitoring.
So returning to the frame challenge, your yuan-ti parents would not just use candling to "discover what kind of child they're going to have" - they would actively choose what kind of child they wanted, and candling would then just be confirmation that development was proceeding as they expected. The implications of this difference for the parents and for the society in general are immense, because this introduces the concept of "designer babies" to a pre-industrial society.