Sarsiz Chauhan

Automating Loan Documents processing for different types and formats; Reducing time to check thousands of loan documents; Using Machine Learning (Random Forest) to create Financial Cost model to detect at risk Financial Centers.

Created a library to convert any kind of money related term($1.5M, One Million Dollar, etc) to numeric ($1,000,000). Published on Python Software Foundation. Data Scraping, saved thousands of man hours by creating and automating the data pipeline processing and storing clean information. Developed a Program to identify Money Related Entities pertaining to Legal Documents with the application of NLP. Trained a binary classification model to classify data acquired from various sources. Creating APIs (in FastAPI, Flask and CherryPy) for the model trained. Establish a data-driven culture with Dashboard reporting, visualization and analytics. Used Deep Learning to make an image captioning model to label the images if they are related to natural calamity or not. Multi-processing was implemented to speed up the task and to improve the utilization of system resources. Socialize different machine learning and data science models with Earthquake, Flood and Remote Sensing Teams. Promoting best practices in Data Management, Analytics and Machine learning.

Developed a POC for Chatbot for HR. Used Tensorflow and Google NMT. Explored RASA NLU for conversational bot. Responsibilities included roadmap, development and implementation of real-time risk and pricing models, maintenance and development of analytical calculation framework.

Written several Scripts in Python to scrap Secondary Data from hundred of websites. Data gathered was cleaned using Pandas, excel and visualized on specific parameters using matplotlib and visualization tools. Used mongodb in backend.