AI Intern — NLP data-extraction pipeline and NER improvement (Sopra HR Software)
Built an NLP document information-extraction workflow using spaCy for 50+ HR templates, focusing on extracting structured data from document text. Improved Named Entity Recognition performance by 25% through a custom spaCy pipeline and pattern optimization. Used Python and regex-based processing to enhance throughput and reliability of downstream extraction outputs. • Developed custom spaCy pipeline components for HR template parsing • Optimized pattern rules to improve entity accuracy • Integrated extraction logic into existing intake workflows for automation • Supported processing of multiple HR document templates for consistent outputs