Key Insights: Built an NLP-based recommendation system that reduces recruiter search time and delivers up to 30% lower cost-per-hire.

About the Client

Experfy is a freelance job platform, incubated at the Harvard Innovation Lab and Deloitte’s primary partner for AI projects. Matching jobs with ideal candidates is both their top priority and the core of their business model.

The Challenge

Recruiters often have to review dozens or even hundreds of profiles to find the right fit for a position. This manual process is slow and resource-intensive, making it harder to connect qualified talent to open roles quickly.

Experfy needed a solution to automatically recommend the most relevant candidates for each job, reducing search time while improving match quality.

Marvik’s Approach

We developed a natural language processing engine to identify candidates with the most relevant skills for a given job posting. Key elements included:

• Benchmarking state-of-the-art models such as Google BERT to ensure accuracy and relevance.
  
  
• Designing a recommendation system capable of ranking and prioritizing candidates automatically.
  
  
• Deploying a scalable architecture built with Python, TensorFlow, Docker, Golang, AWS, and Transformers.
  

The Results & Impact

• Recommended only qualified candidates for each position, improving match rates.
  
  
• Reduced recruiter search time by automating profile screening.
  
  
• Enabled up to 30% reduction in cost-per-hire compared to traditional processes.
  

With the system in place, recruiters can focus on interviewing and selection rather than repetitive filtering, improving both speed and hiring outcomes.

Why This Matters

By applying advanced NLP to talent matching, Experfy can deliver faster, more precise connections between employers and freelancers, turning recruitment into a competitive advantage.