Natural-Language People Search API

Turn a sentence into a ranked shortlist of people.

Describe who you're looking for in plain English. One API call searches 1B+ profiles by meaning, holds your hard filters, and returns a relevance-ranked list.

Query

Backend engineers who built payments or billing infrastructure at an early-stage fintech.

Query

Backend engineers who built payments or billing infrastructure at an early-stage fintech.

Crustdata Semantic Search

Staff Engineer

backend infra · India

Strong

Backend Engineer

payments infra · Portugal

Strong

Intermediate SWE

payment systems · S. Africa

Strong

🎉 We hit $10M ARR!

$600k to $10m in 1.5 years. Our entire GTM is powered by our own APIs.

Data

Delivery Methods

Use Cases

Solutions

🎉 We hit $10M ARR!

$600k to $10m in 1.5 years. Our entire GTM is powered by our own APIs.

Login

🎉 We hit $10M ARR in 1.5 years!

Login

1B+

1B+

profiles searched by meaning, refreshed from 15+ sources

profiles searched by meaning, refreshed from 15+ sources

1B+

profiles searched by meaning, refreshed from 15+ sources

0.03

0.03

credits per result same as filter search, no premium for semantic

credits per result same as filter search, no premium for semantic

0.03

credits per result same as filter search, no premium for semantic

Ranked

Ranked

and tagged strong, possible, or weak on every result, one call

and tagged strong, possible, or weak on every result, one call

Ranked

and tagged strong, possible, or weak on every result, one call

01 · Setup

A sentence in, ranked people out.

Send your query in plain English. Add hard filters only when you have a non-negotiable like location or seniority.

It's the same endpoint you already call. Add one line, nothing to migrate.

Region · United States

Region · United States

Region · United States

Seniority · Senior +

Seniority · Senior +

Seniority · Senior +

Headcount · 1–200

Headcount · 1–200

Headcount · 1–200

Query

built payments infrastructure

Query

built payments infrastructure

Query

built payments infrastructure

Filter = Job Title

Machine Learning Engineer

Filter = Job Title

Machine Learning Engineer

Filter = Job Title

Machine Learning Engineer

Aisha Rao

Aisha Rao

Aisha Rao

Machine Learning Engineer

Machine Learning Engineer

Machine Learning Engineer

Found

Found

Found

Liam Chen

Liam Chen

Liam Chen

Applied Scientist, ML

Applied Scientist, ML

Applied Scientist, ML

Missed

Missed

Missed

Marco Silva

Marco Silva

Marco Silva

Head of Machine Learning

Head of Machine Learning

Head of Machine Learning

Missed

Missed

Missed

02 · The problem

Filters match labels, But people don't fit labels.

The same job title can be written ten different ways and exact filters require you to manually fill each variation.

03 · What it is

Describe the person.
Get the closest matches, ranked.

Our semantic search API matches a profile based on meaning, not just keywords, and puts the strongest fits on top.

Query

built payments infrastructure

Query

built payments infrastructure

Query

built payments infrastructure

Thabo Nkosi

Job title

Intermediate Software Engineer

Work experience

Built and scaled payment systems and billing tools.

Thabo Nkosi

Job title

Intermediate Software Engineer

Work experience

Built and scaled payment systems and billing tools.

Thabo Nkosi

Job title

Intermediate Software Engineer

Work experience

Built and scaled payment systems and billing tools.

"payment systems"

"payment systems"

"payment systems"

matched on work experience not the job title.

matched on work experience not the job title.

matched on work experience not the job title.

How it Works

How it Works

The reading of your query and ranking profiles happen on our end using our proprietary ML algorithm.

The meaning net and the re-rank are ML models, running on the whole database.

The reading of your query and ranking profiles happen on our end using our proprietary ML algorithm.

The meaning net and the re-rank are ML models, running on the whole database.

Your sentence

plain English

Your sentence

plain English

Your sentence

plain English

Crustdata

semantic ranking engine

Meaning net · vector search

Keyword net · exact terms

Re-rank · best on top

Crustdata

semantic ranking engine

Meaning net · vector search

Keyword net · exact terms

Re-rank · best on top

Crustdata

semantic ranking engine

Meaning net · vector search

Keyword net · exact terms

Re-rank · best on top

Ranked results + fit tags

Staff Engineer

strong

Applied Scientist, ML

possible

Ranked results + fit tags

Staff Engineer

strong

Applied Scientist, ML

possible

Ranked results + fit tags

Staff Engineer

strong

Applied Scientist, ML

possible

split

split

split

Semantic search finds the right kind of person. Filters keep the non-negotiables.

The filters gate who is even eligible. The semantic search ranks everyone who clears the eligibility criteria.

1B + profiles

1B + profiles

1B + profiles

whole database

whole database

whole database

Hard filter

Hard filter

Hard filter

must match

must match

must match

Ranked shortlist

Ranked shortlist

Ranked shortlist

scored on meaning

scored on meaning

scored on meaning

Aisha Rao

Aisha Rao

Aisha Rao

Strong

Strong

Strong

Marco Silva

Marco Silva

Marco Silva

Strong

Strong

Strong

06 · Built for builders

One cheap call to Crustdata instead of burning LLM tokens.

If your product's search box accepts natural language queries, route it to one API call instead of paying an LLM to read profiles every time.

Fire it thousands of times a day · no LLM tokens on your side.

Without Semantic

Query

Tokens burned

LLMs

Credits used

Crustdata API calls

Without Semantic

Query

Tokens burned

LLMs

Credits used

Crustdata API calls

Without Semantic

Query

Tokens burned

LLMs

Credits used

Crustdata API calls

With Semantic

Query

Credits used

Crustdata API calls

With Semantic

Query

Credits used

Crustdata API calls

With Semantic

Query

Credits used

Crustdata API calls

07 · What teams build with it

07 · What teams build with it

What teams build with it

What teams build with it

Candidate sourcing

Paste a job description or describe the person, get a scored shortlist you can trust.

Candidate sourcing

Paste a job description or describe the person, get a scored shortlist you can trust.

GTM list-building

Describe your audience in plain English and build outbound lists without manually adding filters.

GTM list-building

Describe your audience in plain English and build outbound lists without manually adding filters.

Market mapping

Find adjacent people and companies that match a background, not just an exact title.

Market mapping

Find adjacent people and companies that match a background, not just an exact title.

Expert-network sourcing

Match by expertise while hard filters exclude anyone at the target company.

Expert-network sourcing

Match by expertise while hard filters exclude anyone at the target company.

Search inside your product

Give your own app a natural-language search box powered by one ranked call.

Search inside your product

Give your own app a natural-language search box powered by one ranked call.

Shrink a huge list

Go from ten thousand people to the few most relevant, already ranked.

Shrink a huge list

Go from ten thousand people to the few most relevant, already ranked.

08 · Is this for you

You're a fit if

You're building natural-language search into a product or agent.

You want a ranked shortlist, not a ten-thousand-row dump.

You need meaning and hard filters in the same query.

You've used search that returns matches you can't trust.

Get your hands on the new way to search for people

FAQ