Yadunandan Huded

Work involves building machine learning models for optimizing Ad-serving, ad-bidding processes and computer vision and NLP APIs to enhance inhouse AdTech products. Built and deployed a ad-optimization model using random forest algorithm which will help reaching the desired views of the digital ads on webpages. Worked on building image search engine using CNN and converted solution to a django API and deployed the API on AWS. Developed image autocrop funcionality based on the user comment, the solution involved using combination on Posenet and NLP.

Worked as a senior analyst for a retail firm and delivered machine learning models. Developed and deployed text classification model leveraging data sets which is obtained by scraping the websites for compititor products. Developed an web application which extracts text information from user uploaded receipts using Tesseract OCR engine and Regular Expression.

Worked as an analyst and delivered solutions primarily in Insurance domain for Australia based Insurance giant & built models in Claim Analytics space. Built a classification model for financial speciality market data using logistic regression and deployed using sql stored procedures. Developed an Anti-fraud model for motor claims data which was available in both structured and unstructured format using random forest model and used Boruta for feature selection. Involved in model validation using statistical procedures and A/B testing for different models developed.

Built multiple POCs and involved in research and development work to optimize existing device driver algorithms. Built a integrated firmware flashing tool, which can used to flash firmware for mobiles and tablets.