Abhinav Tyagi

Developed an ensemble machine learning locally wieghted algorithmic model to predict errors in Logistic billing systems that has high carnality in categorical values and imbalanced data improved the model to achieve an average accuracy of 20%. Build a web scrapping tool for logistic accessorial charges which scrapped US addresses from Bing, Zillow and Yellow-Pages and build a NLP Model using SpaCy which can predict the category and sub-category with an accuracy of 60%. Build website and Flask based API using Sqllite called AddressGuru9 hosted on AWS, which provides scrapped data of US Address to the logistic auditors. Developed an Image Similarity OpenCV model which can club similar images using Imagehash technique and optimised by VP-Trees. Developed hourly report python automation script decreased human effort to nearly zero, which has capabilities to read attachments direct from the outlook mails and create dashboard and send mails with attachment of excel files to the logistics stakeholders.