Utsav Shukla is a Data Scientist at PolicyBazaar with expertise in machine learning and natural language processing. He has developed high-performance NER models and fine-tuned ASR systems for Indic languages. Utsav holds a B.E. in Computer Engineering from Thapar Institute of Engineering and Technology and has previously worked at Skit AI and Qatar Computing Research Institute.
Experience
Data Scientist
PolicyBazaar
Developed over 10 classification and Named Entity Recognition (NER) models leveraging custom BERT architecture. Fine-tuned and deployed the Whisper Large model for Automatic Speech Recognition (ASR) on an in-house dataset comprising Indic languages. Developed a Question and Answer (QnA) system using BERT and T5. Developed and maintained internal tools for detecting data drift and created Python bindings for regular expression matching in Rust.
Machine Learning Engineer
Skit AI (Previously Vernacular AI)
Developed and delivered multi-lingual Intent Classifiers and Entity Extractors for over six clients. Contributed to the development of an in-house automated Machine Learning platform designed to streamline the process of fetching tagged data, training models, storing artifacts, and deploying models on a Kubernetes cluster.
GSoC Intern, GSoC/GCI Mentor
Google Summer of Code
Developed and deployed classifiers capable of identifying Hate Speech and Clickbait. Created a Chrome Extension named 'Social Street Smart' to detect and flag unwanted content. Acted as a project mentor for college students contributing to the 2020 edition of GSoC program.
Data Science Intern
Khushi Baby
Worked on data pre-processing and modeling using health data of the infants and mothers, provided insights to the data with the help of visualizations and descriptive statistics.
Research Intern
Qatar Computing Research Institute
Developed classifiers for detecting bias and factuality of a news media, leveraging audience homophily graph-based techniques and image analysis.
Software Developer Intern
Zuzu AI
Contributed to the development of scalable serverless NLP pipelines aimed at constructing a knowledge-based answering system for customer-facing teams. Involved in the integration of ZuzuAI services with various email and chat platforms.
Education
Thapar Institute of Engineering and Technology
B.E
Computer Engineering
B.E Computer Engineering (CGPA: 8.08 / 10)