Since you're looking for users and not necessarily tweets, you can scrape the company names you're interested in via a spreadsheet, and then append this URL to each unique name: http://twitter.com/#!/search/users/
Just getting rid of duplicate company names should cut down the list some, but it's still going to be a tedious process. At least with the spreadsheet you can add further columns to prioritize your work and go after the companies that are most applicable.
To cut the tedium, Mechanical Turk could then process the results for you fairly quickly.