Portfolio study
Salifort — People Risk Lab
An interactive product that turns an attrition study into signals, scenarios and decisions worth investigating.
This Google Advanced Data Analytics capstone started from a public HR dataset and evolved into a decision product. Instead of opening with raw data, the experience guides the reader through three questions: how large is the problem, where does risk change behavior, and what happens when a scenario changes.
From analysis to product
- Executive view with headline indicators, context and priority findings.
- Exploration of relationships across satisfaction, workload, projects and compensation.
- Risk lab connected to a Random Forest model through a FastAPI service on Cloud Run.
- Resilient experience with versioned aggregates when the API is unavailable and transparent demo-mode labeling.
- Responsive frontend with route splitting, container queries, View Transitions and reduced-motion support.
Limitations
The signals are associations found in this dataset. The application treats prediction as a starting point for investigation, not as an automated decision about people.
Gallery
