Upwork
Scaling the World's Largest Freelancer Marketplace
Upwork, the world's leading freelancer marketplace connecting millions of businesses with independent professionals, partnered with YUPL to address critical scaling challenges and implement next-generation features. Our engagement focused on platform performance, AI-powered talent matching, and enhanced security infrastructure.
With over 18 million freelancers and 5 million clients on the platform, even marginal improvements in performance and matching accuracy translate to significant business impact. YUPL's expertise in high-scale systems and machine learning enabled transformative improvements across the platform.
The Challenge
Operating at Upwork's scale presents unique engineering challenges. The platform processes millions of job postings, proposals, and messages daily, whilst maintaining sub-second response times. The matching algorithm must balance multiple factors including skills, experience, availability, and historical performance to connect the right talent with the right opportunities.
- Peak traffic exceeding 50,000 concurrent users
- Complex matching algorithm requiring real-time processing
- Payment processing across 180+ countries
- Fraud detection across diverse transaction patterns
- Search functionality spanning millions of profiles
Our Solution
YUPL deployed a dedicated team of platform engineers and ML specialists who embedded with Upwork's existing teams. We implemented a comprehensive performance optimisation programme that reduced API response times by 40% through database query optimisation, caching strategies, and microservices refactoring.
Our ML team developed an enhanced talent matching model using transformer-based architectures that significantly improved match quality. The new model considers contextual factors like project complexity, communication preferences, and historical success patterns to surface the most suitable candidates.
Project Gallery
Our work with Upwork touched multiple areas of their platform, from backend infrastructure to user-facing features that serve millions of professionals worldwide.
The Results
Our partnership with Upwork delivered measurable improvements across key platform metrics. The enhanced matching algorithm and performance optimisations have directly contributed to increased user engagement and transaction volume.
- 40% reduction in average API response time
- 28% improvement in talent-job match quality scores
- 35% reduction in time-to-hire for clients
- 60% decrease in fraudulent transaction attempts
- Infrastructure costs reduced by 25% through optimisation
Project Info
Upwork Inc.
Enterprise Engagement
San Francisco, USA
Technology / Marketplace
October 2023
Python, TensorFlow, PostgreSQL
Ready to scale your platform?