Machine Learning Engineer at Infinite Infolab

Architected scalable ETL pipelines with Apache Airflow and Python to ingest sales, inventory, and market data from SQL Server and ERP APIs, improving data freshness and reducing manual extraction effort by 35%. Engineered 200+ temporal and categorical features with Pandas and scikit-learn, including automated data validation and outlier detection to improve model input quality across SKUs. Built ensemble demand forecasting models (XGBoost, LightGBM, Prophet) achieving 18% lower WAPE for 30-day-ahead predictions across 5,000+ SKUs in 12 regional distribution centers. Designed an inventory optimization module using OR-Tools to convert forecasts into replenishment recommendations, reducing stockouts by 28% and excess inventory carrying costs by 22%. Developed Flask-based REST prediction microservices integrated with SAP and Power BI dashboards. Implemented monitoring and retraining workflows with MLflow and GitHub Actions to detect drift and redeploy, cutting update cycles from 3 days to