Sapna Sharma

Data Science Fellow

Fellowship.AI•  May 2020 - August 2020

Healthcare (Deep Learning, Python) • Deployed end to end Wound Tissue Classification App using resnet models with an accuracy of 89% • Scraped more than 800 images from google search engine for database preparation for 7 wound types Virtual Meeting Room Analytics (Computer Vision, NLP, Python, OpenCV) • Developed pipeline for Speaker Diarization by video and audio of a zoom meeting with an accuracy of 50% • Compiled meeting summary based on word frequency and sentence score giving about 50% accuracy • Explored Topic modelling using Latent Dirichlet Allocation, guided LDA for meeting summarization

Teaching Assistant

Northeastern University•  January 2020 - April 2020

● Mentored students in office hours. Graded assignments, quizzes, and midterms. Proof read assignments (15hpw)

Data Analyst

BSNL•  January 2007 - July 2019

• Identified key factors for high latency in performance reporting and reduced the time by 60x by developing an app • Initiated cross-team effort to improve the Key Performance Indicator by reducing overall congestion by 50% • Application to maintain history and print performance report of more than 50 nodes (OCHLA), Application to change tariff of different subscribers to different areas with different time schedule (INDIN). • Application to study and modify subscriber characteristics (ABOIL), An application to store the meter reading of subscribers (TAXIL), Application to highlight the major alarms in real time (ALAIL) Applications developed for KPI maintenance and Fault monitoring • An application to print report and maintain history of fault: FLTRPT • An application to reconcile the data across different technologies in BTS maintenance: DUMPAUDIT