Data Scientist

Role

• Processed 14 million biomedical literature data to prepare company’s Knowledge Graph for biomedical entities and their interactions.
• Worked on developing Extractive Question Answering using SPECTER + FAISS architecture and Entity Normalization and Linking using BioSyn model.
• Boosted the Named Entity Recognition model by an average of 5% F1 Score per entity and improved the multi-label Relation Classification model by 4.2% Micro F1 Score
• Achieved 78.4% Macro F1 Score on a three class Sentiment Analysis model and 55% precision on LDA based topic model for Life Sciences forums and blogs data.
• Developed models for Coreference Resolution, Adverse Event Classification and Irrelevant Sentence Identification.