Is it possible to partially build a pydantic basemodel before validation?

Question

Working on a data access package which handles queries from a database and should return pydantic models.

Here is the Problem:
Most of the models are derived from multiple tables. This means after querying a models table, do not yet have all of the models properties (I first have to query them from tables which are given by foreign keys for example).

Current (bad) solution
Currently I have simply solved this by making the properties Optional and constructing the model in steps. The database queries are set to return dictionaries, so I use model_validate() to instantiate the models. Here is an example:

with self.connection.cursor(dictionary=True, buffered=True) as cursor:
    cursor.execute(
        query,
    )
    result = cursor.fetchone()
    datasets = []
    while isinstance(result, dict):
        dataset = DatasetModel.model_validate(result)
        dataset.format = FormatModel.model_validate(result)
        dataset.scans = self.scan_access.get(
            dataset_ids=[result["dataset_id"]]
        )
        result = cursor.fetchone()

The issue with this is, that in fact these properties are not optional and in the code using the package a lot of asserts are needed to keep the IDE quiet (I am using types, black, ruff,...)

Other solutions I thought of but don't like are tinkering with the results dictionary (adding all properties), before calling model_validate() or creating separate models mirroring the respective tables (both seem like a bit too much hassle tbh).

So the question is simple: Is it possible to construct the model partially and do the validation only when it is completed?