Learning the world of finance and working on In-house CIBIL Score Card modeling, Customer Early Warning Signals models on different products and much more.

Developed a production-grade Spacy NER model based on BERT architecture, automating entity extraction from deed documents and reducing manual effort for the BPS team by 75%.Implemented a streamlined workflow utilizing Google Vision API for document processing, custom Spacy model for entity recognition, and Google Vertex AI with custom containerization for model deployment.Created an efficient solution for extracting critical financial data from extensive PDF document datasets… Show more Developed a production-grade Spacy NER model based on BERT architecture, automating entity extraction from deed documents and reducing manual effort for the BPS team by 75%.Implemented a streamlined workflow utilizing Google Vision API for document processing, custom Spacy model for entity recognition, and Google Vertex AI with custom containerization for model deployment.Created an efficient solution for extracting critical financial data from extensive PDF document datasets, resulting in a 60% reduction in user processing time and improving overall efficiency and accuracy.Tech Stack - GCP, Vertex Ai, Google Vision API, SpaCy with BERT, Pandas, Sk-Learn, Regex, Transformer Show less

Developed ETL pipelines to efficiently load data from EDW to multiple applications, reducing processing time by 50% and increasing data accuracy by 30%.Developed efficient ETL processes to extract data from multiple systems and load it into the EDW layer, resulting in a 30% reduction in data loading time.Implemented automation of job scheduling using Redwood scheduler, resulting in on-time delivery of critical data to downstream systems with a success rate of… Show more Developed ETL pipelines to efficiently load data from EDW to multiple applications, reducing processing time by 50% and increasing data accuracy by 30%.Developed efficient ETL processes to extract data from multiple systems and load it into the EDW layer, resulting in a 30% reduction in data loading time.Implemented automation of job scheduling using Redwood scheduler, resulting in on-time delivery of critical data to downstream systems with a success rate of 98%.Collaborated with cross-functional teams to gather business requirements and ensure successful project delivery, resulting in a 95% satisfaction rate from stakeholders. Show less