Sarath Lavu
@sarathlavu
Data Scientist at HiLabs
Bangalore, India
Sarath Lavu is a Data Scientist at HiLabs with a background in Economic Sciences from IIT Kanpur. He has expertise in developing LLM-based pipelines, multi-class classifiers, and optimizing data processing workflows. His technical skills include Python, PySpark, and AWS, complemented by strong analytical experience in econometrics and time series analysis.
Experience
Data Scientist
HiLabs
Developed a question to report generation pipeline using LLMs deployed on sagemaker to fetch relevant data & plot graphs. Implemented a multi-class classifier to map raw columns present in data. Achieved an accuracy of 98%. Enhanced data processing pipeline efficiency by up to 8x through utilization of pandas’ built-in filtering, optimization of data types, and targeted performance enhancements. Designed and implemented the automation of data extraction, cleaning and parsing process, leading to a 70% reduction in manual work. Restructured and optimized the code base to shift from windows to linux aws cloud resources thereby reducing the costs by 50%.
Data Science Intern
HiLabs
Implemented a classifier model to predict the credibility of a healthcare related webpages. Achieved 90% F1 score using LightGBM classifier. Developed scripts to extract, clean and parse unstructured data in various formats such as html tables, pdfs from more than 10 different health insurance websites.
Education
IIT KANPUR
MS
Economic Sciences
Dual Degree
IIT KANPUR
BS
Economic Sciences
JUPITER JUNIOR COLLEGE
BIE, A.P