Resume Parsing and Semantic Search Project
Problem
- Company receives 3 million job applications each year - with such a high volume, a thorough review of each resume/profile is challenging and qualified applicants may be overlooked.
- The resume review process is highly subjective, depending heavily upon the HR specialists' area of expertise and level of experience.
Action
- Implemented a process to import resume raw-data from various recruitment application portals and extract useful information
- Stored parsed resumes from various sources in a centralized database for querying and ranking
- Capability to Query and Rank resumes based on predefined criteria and take actions
- Capability to Search and score resumes based on semantic similarity between words and not just text
Tool Stack
- Python, Flask, Gunicorn, Tensorflow, MLFlow, MS Azure, .net core, RabbitMq, Elasticsearch, MS SQL
89
%Reduce data entry time
85
%Boost in candidate experience