Blog

Notes from the team building SignalGrid.

Modeling decisions, engineering tradeoffs, and the occasional unpopular opinion about how this category should work.

  • 7 min readDraft

    The case for cross-vertical entity graphs

    Most signal-intelligence tools resolve entities inside their own vertical and stop. We argue the interesting predictions live across the seams.

    Priya Ramanathan · Head of ModelingRead
  • 6 min readDraft

    Why we don't trust survey-based Q-scores

    Annual phone surveys of a few hundred respondents cannot keep up with how fast public attention now moves. Here is what we use instead.

    Daniel Okafor · Lead, Celebrity EquityRead
  • 9 min readDraft

    How we built the kynetranews ingest layer

    Forty-nine RSS feeds, six social platforms, four search products, and one knowledge graph. Notes on what we learned shipping it.

    Marcus Yi · Engineering Lead, IngestRead
  • 8 min readDraft

    Calibration matters more than accuracy in flop prediction

    A 90 percent accurate flop predictor that is over-confident is worse than an 80 percent accurate one that knows when it does not know.

    Priya Ramanathan · Head of ModelingRead