Soham Nanavati

With over two years of experience building and deploying machine learning solutions in a large-scale manufacturing environment, I bring strong expertise in developing data-driven systems that optimize industrial operations and deliver measurable business impact.

In my current role at PGP Glass Pvt. Ltd., I design and deploy end-to-end machine learning and data engineering solutions on Databricks to support complex production processes. I have worked extensively with large-scale industrial datasets, developing predictive and optimization models to improve operational efficiency. A key example is a Kernel Density Estimate (KDE)-based optimization model that identified optimal furnace temperature ranges, significantly improving Pack-to-Melt efficiency and delivering verified savings of over 21 million LKR.

I have also built scalable data pipelines and analytical systems that integrate multiple data sources to enable decision-making at scale. My ROIC analytics platform consolidates sales, EBITDA, and capital data to provide SKU-level profitability insights for leadership. Additionally, I developed a full-stack AI platform that combines real-time data processing with machine learning and intelligent querying, enabling engineers and operators to analyze production performance and identify root causes of defects.

My technical expertise includes Python, PySpark, SQL, Databricks, Delta Lake, and MLflow, along with strong experience in model development, feature engineering, and deployment in production environments.