Problem - A global enzyme provider needs to promptly recommend the right products and dosages and estimate costs for its customers during sales meetings to enable buying decisions. The enzyme products are used in foods and beverages, detergents, biofuels, and more.

Approach - Utilize ML models to predict the qualities of the customers' products (baking products, washing detergent, biofuel, etc.). An optimizer recommends the optimal enzymes and dosages to achieve the best qualities and lowest costs.

Results - A web application was delivered. The client's sales teams or customers can log in, identify recommended enzymes with estimated costs, and possibly make purchases.

Data - Enzymes, product types, product recipes, product conditions, geographical areas, and time.

Tools - Database: SQL; Data science: Python, Scikit-learn, Jupyter, ML Flow, AWS Sagemaker; Web app: Python Pyramid, React JS.

Team - A data scientist, a backend and cloud developer, and a frontend developer.