Skip to content
Trusted Media Brands2026Live

StockPull

CLIP-scored image search that lets editorial researchers query four stock and DAM libraries at once instead of four separate logins.

CLIPGettyAdobe StockIAP
Concept mockup of a dark image-search interface: a sidebar reading StockPull above Search, Collections and History, one wide "Search for images…" field, a row of filter chips for Photos, Orientation, Colour and Aspect Ratio, and a grid of unlabelled landscape and architecture photographs. One tile is outlined in vivid orange with a partly filled orange relevance bar beneath it.
Concept mockup — one result grid, with the top CLIP match lifted out and ranked on visual relevance.

The Problem

Editorial photo researchers had to search Getty, Adobe Stock, Woodwing, and OrangeLogic separately: four logins, four keyword boxes, one image. Keyword search also misses images that look right but were never tagged right.

What I Built

StockPull is an internal search tool gated behind a single Identity-Aware Proxy login that queries all four libraries at once. A CLIP visual scoring layer ranks results by what an image actually looks like, not just its metadata, so a researcher can search by concept instead of guessing the right keyword.

Selected Work

Concept mockup of a "Querying sources" panel: four stacked source-library rows, each with a partly filled progress track and a grey "Querying" status, gathered by an orange connector into a single merged "Ranked results" row whose count slot shows an em dash instead of a number.
Concept mockup — four libraries queried in parallel behind one Identity-Aware Proxy login, fanning into a single ranked list.
Concept mockup of a visual-similarity view: a large reference photograph of a fog-covered pine ridge on the left, and on the right a "Visually similar" row of four closely matching misty forest images, each above an orange relevance bar that steps shorter from left to right, under the caption "Ranked by visual similarity".
Concept mockup — the CLIP layer ranks neighbours by what an image looks like, not by how it was tagged.

How It Works

01IAP-gated query
02fan-out search across Getty, Adobe Stock, Woodwing, OrangeLogic
03CLIP visual scoring
04ranked results

Outcome

Researchers get one search box instead of four, with results ranked on visual relevance rather than tag luck.

My role

Creative Technologist at TMB — designed the pipeline, built it, and shipped it to production.

Stack

Google Cloud PlatformIdentity-Aware ProxyCLIP visual embeddingsGettyAdobe StockWoodwingOrangeLogic