
@jnunemaker Might not be helpful for live analytics, but a pixel to flat files via a log then separating compute from data (eg querying Parquet files from disk or S3) is pretty cool.
Our biggest challenge now is filtering automated/scraper traffic to better trust the data.
English





