Sonal Gupta

Item code Identification for Airport Customs: Pre-processing of historical data containing different languages; Building a KNN based approach to identify first level of code for the query item. Resume Data Parsing: Improved email extraction flow by 5 percent using fuzzy regex; Used NER model results along with document structure and custom algorithms to find current organization and classify a resume as fresher or Experienced; Built a classification model to segregate resumes belonging to different technologies. PDF document parsing and Segmentation: Extracting and segmenting data from unstructured pdf documents into a structured json format; Built a Text Classification model using XgBoost to determine whether a text in a pdf is heading or not; Incorporated image information in table data obtained using Textract and Camelot; Identified and resolved bugs within the existing pdf data curation pipeline. Custom Wake Word Detection and Voice Assistant: Built a CNN based speech classification model for wake word detection; Integrated Mlflow in training pipeline; Built and dockerized a web application for real-time testing; Built a POC for voice based search using Amazon Lex. Traffic Monitoring System: Built custom algorithm to use optical flow and object detection model to track movement of vehicles on crossroads and calculate approximate speed of vehicles.

Banglore Remote

Gorakhpur