.

if you have a representative sample of people qualifying for the jobs together with their characteristics you might consider matching algorithms based on bipartite graphs (with edge strength between two clusters being the number of units common to two clusters) , hungarian algorithm for instance

.

the way your question is formulated seems to indicate that this is not the case, that you have performed two k-means on independent data sets (people characteristics on the one side, job characteristics on the other side)

then ... there is no miracle and no matching algorithm without injection of more knowledge

for instance, job description and people education should share common descriptions etc : the matchnig could be made by extracting relevant informations common on both sides, building cluster representations on those informations and compute a matching score between clusters

i do not think you'll hack through this without a little programming ... :(

.

In your problem isn't it more convenient and more flexible to formulate a scoring scheme to rank each record (the eligibility degree)?

First, you should define a similarity function between a person and a job. Then, for each centroid of your people groups, compute the similarity with the centroids of the job groups and sort them for increasing distance. This is like a similarity join.

Sir, I think I have a solution for your problem based on generating some functions that can correlate the variables. We can discuss about the problem face to face if you have no problem. (Thank you)

Thank you very much. @ Fabrice Clerot Ali Rezaee Rodrigo Paredes.