Projects

Live - MVP

LinkedCulture

Unified Cultural Search

Search Harvard Art Museums, the Metropolitan Museum of Art, and Smithsonian Asian collections simultaneously through a single vector search interface. Built on Nomic embeddings and Qdrant. No LLM in the loop. Pure semantic similarity at scale.

Pilot Fit

A useful pilot would index one collection or a small cross-institutional set, then evaluate search quality against real research, education, or public discovery tasks.

Who It Serves

Museums, archives, libraries, researchers, digital humanities teams, and cultural heritage organizations exploring semantic discovery.

Themes

Semantic searchNomic embeddingsQdrantOpen collections

Shared Process

From fragmented inputs to usable outputs.

01

Ingest collection records

02

Generate embeddings

03

Index in Qdrant

04

Search by semantic similarity

LinkedCulture

A research platform for cultural collection discovery: unified vector search across major open-access collections, without an LLM in the retrieval loop.

I support museums, archives, and cultural institutions with digital infrastructure that empowers discovery. Less manual friction, more time for storytelling, stewardship, and shared knowledge.

More
Enriched Metadata
Shared Infrastructure
Open Standards
Searchability
Insight
Less
Manual Tagging
Data Silos
One-Off Tools
Fractured Systems
Opaque Workflows

Workflow: From Raw Data to Public Insight

Ingest

Museum APIs and collection exports

Systems:
Harvard, Met, Smithsonian

Embed

Convert records into semantic vectors

Systems:
Nomic embeddings

Index

Store and query similarity at scale

Systems:
Qdrant

Evaluate

Compare cross-collection discovery quality

Systems:
Search logs and result review

Discover

Search by concepts, materials, symbols, and meaning

Systems:
Next.js interface

LinkedCulture System Relationship Diagram

This diagram visualizes how collection metadata, embedding generation, vector search, and the public interface coordinate in the prototype.