Founder & CEO
Alex was previously at Meta (then Facebook), working on recommendation and location systems, and Jane Street, working on options trading systems.
His first experience with computer vision was at UPenn, where he conducted research on how to use GPUs to find stars. He now finds himself knee-deep augmenting images, tuning parameters, and optimizing inference pipelines most of the week.
Dragoneye's video API now returns objects tracked across the entire video - each with its category and attribute classifications, timestamps, and bounding boxes - unlocking instance counting and spatial-temporal reasoning.
Use the Dragoneye MCP server to let a coding agent build a zero-shot object detection model and wire it into your app — from a stubbed UI to a working vision feature in about ten minutes.
Let coding agents create Dragoneye models and wire image and video detection into your app.
Go beyond bounding boxes: extract structured attributes on detected objects, and build models conversationally instead of writing definitions by hand.
Learn the most common bounding box formats used in computer vision, including COCO, YOLO, and Pascal VOC. This quick guide explains how bounding boxes represent objects in images, the key format differences, and how to choose the right one for your object detection projects.