Business Objective / Goal
To build an ML-driven predictive pricing engine that could accurately forecast monthly supply, demand, and pricing for different food product categories, improve decision-making, and reduce time and cost spent on manual estimations.
Solutions & Implementation
- Developed a Machine Learning-based price prediction engine on Amazon SageMaker for real-time estimation of optimal product pricing.
- Integrated the solution with a CI/CD pipeline using Amazon CodePipeline and CodeCommit for seamless model retraining and deployment.
- Cleaned and engineered data to identify key pricing factors across commercial, retail, and international customer segments.
- Deployed the entire workflow using Docker, Lambda, and API Gateway for scalable and serverless execution.
- Hosted services on AWS EC2 with APIs exposed via Flask for model interaction and consumption.
Major Technologies Used
- Amazon SageMaker – ML training and deployment platform
- Amazon CodePipeline, CodeCommit, ECR – For CI/CD orchestration
- AWS Lambda, EC2, API Gateway – For serverless and scalable API services
- Python, Flask, Docker – Core language and application framework stack
Business Outcomes
- Automated Optimal Price Estimation Enabled accurate price prediction across all food categories, reducing manual effort.
- Improved Forecasting Efficiency Used ML models to augment and refine monthly supply-demand predictions.
- CI/CD Enabled Retraining Pipeline Deployed a flexible retraining mechanism with AWS CodePipeline for scalable model updates.
- Cost and Time Savings Through Automation Replaced manual market forecast processes with a data-driven, ML-based system.
