Overview
Ceramic manufacturing at CoorsTek involves layered material histories — each part's composition traces back through a series of raw materials, intermediate formulations, and processing steps. Manually reconstructing a material's full history and calculating its formulation values required navigating multiple database tables and performing unit-conversion calculations by hand, costing researchers roughly 30 minutes per query.
This project replaced that manual process with a programmatic graph model and a researcher-facing web application.
What I Built
SQL Data Layer
Wrote a program to query the production SQL database and calculate the formulation of each material in a project — including unit transformations so all values are expressed in consistent terms regardless of how they were originally entered.
Directed Acyclic Graph Model
Developed a generic, object-oriented implementation of a directed acyclic graph to represent material histories. Each node holds a material and its calculated formulation; edges encode the processing relationships between them. The graph structure makes it straightforward to traverse a full history, isolate a sub-chain, or compute derived values at any point in the lineage.
Streamlit Research App
Built an interactive Streamlit web application so R&D researchers can query and visualize material histories without writing any code. Researchers select a project, and the app retrieves the data, constructs the graph, and displays the formulation at every node — saving approximately 30 minutes of manual calculations per query.