• Curated a labeled medical dataset constituting ~16k samples using selenium and rule-based NLP techniques.

• Developing an end-to-end Named Entity Recognition (NER) pipeline for Medical unstructured data using AllenNLP toolkit.

Teaching assistant for CSE 352 Artificial Intelligence course

• Devised machine learning (ML) based pipeline for enhancing data quality pertaining to issue management. This involved developing different ML models for imputing missing values and identifying mislabeled samples. Lastly, we leveraged ensemble learning to learn from multimodal data which improved the model’s performance from 76.2% to 80.3%.

• Developed a ML-based semantic search engine by generating synthetic queries based on Sentence-Bert embedding.

• Developed an end-to-end Sentiment Analysis and Named Entity Recognition (NER) models using BERT variants and deployed the trained models using Docker Container on AWS.