Hundred

Hundred is an automated inspection system that catches sub-millimetre cosmetic defects in consumer electronics parts before they hit the assembly line. The aim is to cut factory waste and prevent line-down issues that can turn a finished device into scrap.

Product designs now change faster than traditional vision models can be retrained, so factories still fall back on people checking parts under microscopes — a process that reportedly misses around 15% of defects. With defects costing U.S. discrete manufacturers an estimated $45 billion a year, the economics of manual inspection are harder to defend.

Erica saw the problem up close designing hardware at Meta, Google and Magic Leap. She met Imani at Columbia's admitted-students day in 2017 and later connected with Iris through mutual friends in the Bay Area, forming the team around a simple conviction: manual inspection shouldn't exist in 2026.

The founding team pairs hardware design experience with serious engineering depth — Imani spent five years at Twitch building software handling billions of requests a day, and Iris moved from quant work at Citadel to researching machine learning on 3D meshes and structured data at Imperial College London. That mix of scale software and structured-data ML maps closely onto the inspection problem.