Dataset -> S3 -> EMR (Hadoop MapReduce) -> S3 -> Flask API -> React

A CS5224 cloud project: count followers and followees for every user in a 2.4-million-edge Twitter graph, then serve the counts on the web.

I wrote a single-pass MapReduce job for Hadoop Streaming on AWS EMR, kept inputs and outputs in S3, and served the results through a Flask API and a searchable React table.

The pipeline turns 2,420,766 edges into counts for 81,306 users; the most-followed has about 8,660 followers.