Harsh Agrawal, Ph.D.
About Harsh Agrawal, Ph.D.
Harsh Agrawal, Ph.D., is a Data Scientist at Noodle.ai in Bengaluru, India, where he focuses on predictive maintenance in steel manufacturing. He has a strong background in data analysis and algorithm development, with previous experience at AMOLF and the Indian Institute of Technology, Madras.
Work at Noodle.ai
Harsh Agrawal has been employed as a Data Scientist at Noodle.ai since 2021, based in Bengaluru, Karnataka, India. In this role, he has focused on delivering anomaly detection solutions within the Asset Health Artificial Intelligence vertical, particularly for predictive maintenance in steel manufacturing plants. He has developed a time series clustering and scoring algorithm that identifies anomalies, contributing to significant cost savings by minimizing unplanned downtime. Agrawal has also led weekly client calls to gather use case requirements and explain model training outcomes to various stakeholders.
Previous Experience at AMOLF
Prior to joining Noodle.ai, Harsh Agrawal worked at AMOLF in the Amsterdam Area, Netherlands, from 2016 to 2021. He served as a PhD Researcher and Junior Scientist, where he focused on nanotechnology and completed his Doctor of Philosophy (PhD). Additionally, he was a member of the international panel in the ‘Works Council’ of AMOLF from 2019 to 2021, contributing to organizational governance and decision-making.
Education and Expertise
Harsh Agrawal holds a Doctor of Philosophy (PhD) in Nanotechnology from AMOLF, obtained between 2016 and 2021. He also earned a Master of Science (M.Sc.) in Materials Science from the University of Stuttgart from 2014 to 2016. Earlier, he completed a Bachelor of Technology (B.Tech) in Metallurgical and Materials Engineering at Visvesvaraya National Institute of Technology from 2009 to 2013. His educational background provides a strong foundation in materials science and data analysis.
Data Analysis and Model Development
In his professional roles, Harsh Agrawal has handled extensive exploratory data analysis, refining and creating features to support model training with large datasets ranging from 5 to 40 million rows. He has owned multiple end-to-end use cases for components like rollers and brakes, successfully deploying over 15 models. His contributions include creating functionality for data pre-processing across multiple time zones and developing evaluation metrics for deployed models.
Mentorship and Collaboration
Harsh Agrawal has demonstrated a commitment to mentorship by guiding a student in developing an exploratory data and failure analysis module. This initiative significantly reduced the effort required for exploratory data analysis by at least 50%. He has also collaborated with data scientists and maintenance engineers, clarifying domain knowledge and enhancing communication between technical and non-technical stakeholders.