I understand the project, and I've done this in the past, many many times.
My favorite approach is a mix of Perl and Expect (to connect to the Whois services), and I believe I can redo it pretty fast. (My code is pretty old, dating to 96-99, if I still have it in some obscure backups).
One pitfall though in using the public Whois services, is that now most of the Whois lookups are IP-limited (depending on the registries it goes from 10 to 10000 lookups a day), so processing 100s of thousands per day is practically impossible using the public whois services, although something can be figured out with multiple IPs to spread the workload (depending on your daily goals).
I suspect using a third-party API service from some registrar would help break this limit barrier; but it would certainly incur additional costs (service membership and/or cost per X lookups).
I have a strong experience in these matters; having written my personal Whois spiders (before there ever existed the limit checks... might be my fault, sorry), limit checks are not a problem for me though; I've worked with that on many spider projects (and still do).
20 years+ experience on Unix (BSD, Linux, Solaris), and I'm pretty confident I can have it running the same on Mac or any unix flavor.
Fluent in french, english and spanish, I can also handle international character sets.
Your budget is quite small though; I would prefer bidding 500-800$ on such a project. But I can make do with it, if you don't require too much post-production support.