Case studies/Tarjama / Ureed

HR-Tech

Tarjama / Ureed

Built the NLP-powered candidate matching behind the Middle East's largest freelancer marketplace, in 2020.

ClientTarjama / Ureed
IndustryHR-Tech
Duration18 months
Year2020, 2021

The challenge

Ureed was transitioning from a translation platform into a multi-category freelancer marketplace. Matching freelancers to jobs, onboarding new freelancers, reading competitor platforms — all manual. None of it scaled, and the business wanted to 10x its freelancer base.

What we built

1

Candidate parsing automation

A custom NLP pipeline, built in 2020 before ChatGPT existed, that parsed freelancer applications, extracted skills, experience, specialties, and structured them into the platform taxonomy. Zero manual data entry for onboarding.

2

Freelancer-customer matching engine

Automated matching by skills, availability, past performance, and project requirements. Hours of human coordination compressed into seconds.

3

Skill and specialty assignment

Profile and sample analysis that auto-assigns and updates skill tags. Marketplace stays organized without human curation.

4

Competitive intelligence automation

Distributed scraping of Upwork and other marketplaces, pattern detection on pricing and demand, and automatic feed into Ureed's product and pricing strategy.

5

Full platform migration

Sole engineer on the backend migration from translation-only to full multi-category marketplace. Eliminated N+1 query patterns, integrated Algolia and Checkout.com, built the timesheet system.

ROI

Results

Zeromanual parsing on freelancer onboarding
Secondsto match (from hours of human work)
$20M+raised by Tarjama across Series A rounds
18 monthsas core engineer across the transition

Outcome

The automation infrastructure let the marketplace scale its freelancer base and customer throughput without proportionally scaling operations. AI-powered parsing, matching, and intelligence transformed a labor-intensive marketplace into a machine that ran itself, built years before the rest of the industry caught up with AI.

Technologies

PythonDjangoCeleryPostgreSQLCustom NLP modelsAlgoliaCheckout.comDistributed scrapingREST APIs
"

They built NLP-powered matching infrastructure for us in 2020, before the rest of the industry had a word for it. Years later, it is still the engine underneath the marketplace.

Marwan Abdelaziz
Marwan AbdelazizFormer CEO, Ureed

See more work, or start yours.

All case studies, or tell us what you want automated.