one for single cv: it has to be defined on the basis of requirements of templates, in particular: europass (link), moderncv (link), R cv (https://ropensci.org/blog/2019/01/10/vitae/); \TEX candidate templates are: awesomecv hyndman moderncv twentyseconds; table candidates are: personal_data, employers, places, jobs, publications, study, affiliations, referee; some of these might be combined, adding a category, e.g.affiliations and referee can be a cat in jobs
one for organizations, with multiple persons, different sectors (make cv for specific sectors only)
the first can be derived as a view from the second, so the generation scripts can be used without modifications
for translations, two cases:
repeated items (e.g. "Scientific publications"): use ENUMs in the table, then in the script replace with the appropriate translation
non repeated text (e.g. the description of each job): have multiple columns with the content in different languages, e.g. job_en, job_it, etc.