They are ordered by percentage descending, so you can quickly understand which is the 2nd most populous, 3rd most populous, etc. The ethnicity name is right by the percentage so you can quickly grok it, even though the colors don't match across pie charts.
Those are just "thumbnail" ratios, taken by dividing Amazon's percentages off the chart; we don't have enough information to infer the exact ratios, of course. But taking the case of gender, if we make a reasonable guess of a 4:1 ratio of workers-to-management, we can readily pop out a 60-40 split (male-female) for non-management workers, compared with the 75-25 split for management workers.
We still don't have enough information to conclude evidence (from these charts alone) that Amazon has systematic problems promoting women or non-whites to management (we'd need more hard data, and some basis for comparing with other companies). But the point isn't that we've caught Amazon at something, or that the charts are lying; it's rather that the charts are meaningless. They're simply eye candy; just don't really convey anything one way or another about the fairness of Amazon's hiring practices, or about anything else really.