Himanshu Kumar

Developed a predictive risk assessment model using XGBoost for Paytm's Cancellation Protect insurance product. Engineered a collections optimization model utilizing a two-tower neural network architecture. Implemented a base allocation strategy to assign credit cards to customers using linear optimization.

Developed multiple multi-class classification models using XGBoost to classify user reviews. Established streamlined data pipelines and real-time dashboards using Tableau and MySQL. Created dashboards with Grafana to automate extraction of key metrics for mobile application portfolios.

Developed a classification model using XGBoost to identify action items from meeting transcripts. Developed a semantic search engine for querying internal organizational documents leveraging ElasticSearch and BERT embeddings.

Developed a multi-class classification model using Random Forest to categorize customer reviews. Developed a cost-efficient translation memory system using BERT embeddings. Developed an NER model to identify transaction-related entities from documents using Stanford NER and OCR.