Case Study Deskcasestudydesk.com
Recruitment / job searchSourced

Case Study: How DirectlyApply made job search 400x faster with SingleStore vector search

DirectlyApply Case StudySourced & dated by Case Study Desk
Key facts · TL;DR
Company
DirectlyApply
Industry
Recruitment / job search
Challenge
MongoDB and Elasticsearch layers degraded and grew costly as the platform scaled.
Headline result
400x faster query performance with built-in vector search

Key results

400x
Faster query performance
6 ms
Query time
Down from 2,600ms
$100,000
Annual cost savings
80%
Of platform on SingleStore

The challenge

DirectlyApply's job-discovery platform relied on MongoDB and Elasticsearch layers that degraded in performance and grew costly as it scaled. Rising query times forced client-side processing that negatively impacted the user experience.

The solution

DirectlyApply consolidated onto SingleStore Helios, using its built-in vector search and DOT_PRODUCT similarity to match job titles against standardized ISCO occupation data.

We've been dreaming about doing this for four years. Now with the help of SingleStore, we can accelerate helping job seekers find new jobs.

DB
Dylan Buckley
Co-Founder, DirectlyApply

The results, in context

DirectlyApply reports 400x faster query performance, with Elasticsearch queries that took 2,600 milliseconds now returning in 6 milliseconds. It cites $100,000 in annual cost savings and now runs 80% of its platform on SingleStore Helios.

Products used

SingleStore SingleStore HeliosSingleStore SingleStore vector search