Amazon.com is hosting a shindig tonight at the W Hotel in Seattle where it will choose the best startup using its Web services.

Candidates were winnowed down to the seven that are presenting to a panel of judges this afternoon. A winner will be announced tonight and receive $50,000 cash, $50,000 in Amazon Web Services credits and the offer of an investment from Amazon.

Here are the finalists:

-- Brainscape is a free and open-source "database for resting state functional connectivity studies" developed by the Neuroinformatics Research Group at Washington University in St. Louis.

-- Commerce360 is a search-engine optimization company based outside of Philadelphia. Its key technology is "ClickEquations" software for optimizing search campaigns; clients include Comcast and The Franklin Mint.

-- Justin.tv is a live-video portal based in San Francisco but run by a group of former Seattlelites -- founder Justin Kan and pals Michael Seibel and Emmett Shear.

-- MileMeter is a Dallas auto insurance company preparing to offer insurance "buy the mile," selling miles in advance "so people who drive less pay less for insurance."

-- Ooyala is a high-definition Web video delivery and advertising platform started by a group of ex-Googlers, based in Mountain View, Calif., of course.

-- UserTesting.com offers inexpensive ($19) online usability testing of Web sites. It's also based in Mountain View.

-- WeoGeo is a Florida company building "a one-stop marketplace for mapping," giving surveyors, engineers, cartographers and scientists a place to store, search, exchange and sell map products.

I haven't heard their pitches yet, but here's my early, shoot-from-the-hip take.

Web video is getting crowded, but Justin.tv has hometown sympathy, the groovy factor and reaches the hot demographic. Ooyala has huge ambitions, polish and the Valley buzz factor, especially since everyone's been wanting to see what sort of companies rich ex-Googlers will start (do they even need the $100K?).

Brainscape is fascinating but it's not going to be as widely used as the others and won't use as many AWS services -- would you like an MP3 with that scientific paper? I wonder if Amazon will give a break to scientific/public interest projects built with its services.

Commerce360 may be the stuff, but SEO is getting crowded, it's nichey and the company seems a bit regional.

MileMeter is clever but is it a gimmick or a revolution? I can't imagine an insurance company would be able to replace traditional risk factors with mileage. I also wonder if consumers are ready to start buying auto insurance in blocks of time, like phone card minutes; the ones who want sporadic, bargain coverage may not be best customers for an insurance company.

UserTesting.com is a great idea. I wonder if it will be acquired (or imitated) and offered as a feature by a hosting service. OfficeLive, perhaps? But it probably won't showcase the huge scaling capabilities of AWS.

WeoGeo's the real standout to me, given the intense interest in location-based services and their underlying GIS technology. It also has really slick tools for searching and shopping that may catch the eye of Amazon executives doing the judging.

We'll see who wins in a few hours.