Nishan Jain

Machine Learning Lead

Canwill Technology•  March 2019 - Present

Working on a data breach application whose goal is to reduce the time taken by reviewers to identify users whose data has been breached by looking at documents that are uploaded in a large quantity. Machine learning concepts applied are: Image classification to classify images such as passports, death certificates, medical forms etc using Fast AI. Email conversation threading and finding inclusive emails from a large set of emails. Document classification to classify documents such as forms, invoices, mortgage, bills etc using BERT Classifying if the email is ham or spam using BERT Identifying signatures and headers from an email using CRF model Deduplication of millions of entities Building information extraction pipeline using coreference resolution, Stanford open ie, NERs, relation extraction and POS Tagging