Hi,
As you have probably already figured out, work tables are used for temporary storage of data so that it can be accessed in other parts of your data job or process job. It can't be used as input for a SQL statement - at least not directly. It would be possible to read values out of a work table and store the values in macro variables that could be used as parameters to an SQL statement in other steps in your job.
If your goal is to read data into your job from a csv file to then subset it in some way, you could read it in directly, skip the work table step, and use the expression node to select/filter the rows that you are interested in.
Ron