Anyone have suggestions of sources for profiles?
The method for the MVP is to source profiles from Linkedin, where searches on people who are 1st or 2nd degree connections generate the following numbers.
- 2500K/60K: content writing, especially for grant proposals.
- 2500K/600K Python programming, especially for machine learning / data science.
- 100K UX design, especially for problem solving.
- 30K/5K Japanese / English interpreting, especially for financial business.
I wrote a program in Python a year ago to run on Selenium, enabling login to my Linkedin account, and then on a right click, save profile information for any individual from their page, using a combination of Linkedin API and XPATH. In combination with search results as identified above, this method accelerates (without automating, which violates the TOS, subject to appeal of the HiQ lawsuit resolution from the 9th Circuit Court last fall) collection of relevant profiles.
From there, there is a lot of work to generate matching. For the purposes of the MVP, we just plan to use basic ML clustering, possibly with GraphQL access to data in a Cloud Firestore document environment, to pull relevant “smart connections.”
This is the third of three core problems that Marmalade AI is intended to solve: 1) how can I go to meetups and network more efficiently with the right people for me, 2) how can I keep my network up-to-date on me, and 3) how can I get access to “smart connections” I never would find on my own? For the Y Combinator Startup School MVP, even just the third problem seems pretty big when you get closer to it.