MOHANREDDY KAMBHAM

• Designed and deployed end-to-end production ML systems with monitoring, retraining, and optimization.
• Improved fraud detection F1-score by 12% and increased AUC-ROC from 0.82 to 0.89.
• Built real-time inference services handling 1.5M+ requests/day with sub-50ms latency.
• Processed 500M+ records (~1.2TB) using Spark distributed pipelines.
• Implemented CI/CD, drift detection, and automated retraining pipelines.
• Delivered enterprise AI initiatives including LLM workflows and RAG-based solutions.

Developed an end-to-end machine learning solution to predict product demand and optimize inventory planning using historical sales data, seasonal trends, promotions, and external market factors. Built advanced forecasting models (XGBoost, Random Forest, Prophet) with automated retraining pipelines to improve prediction accuracy. Designed an intelligent reorder recommendation engine that minimized overstocking and prevented stock-outs.

Implemented interactive business dashboards for real-time KPI monitoring, demand trends, and inventory health, enabling faster decision-making for operations teams. Improved forecasting accuracy by 35%, reduced stock-outs by 28%, and lowered excess inventory costs by 18%.