Back to projects
Data Analysis

Supply-Chain Stress Dashboard

Real-time analytics dashboard tracking supply-chain stress indicators with automated data pipelines and interactive visualizations.

PythonFastAPIStreamlitSQLAlchemyAlembicPostgreSQL
GitHub

Overview

A full-stack analytics application that aggregates, processes, and visualizes supply-chain stress indicators in real-time. Built to help analysts monitor disruption risks across multiple data sources through a single interactive dashboard.

Problem

Supply-chain analysts often juggle multiple data sources—shipping indices, port congestion metrics, commodity prices—spread across different tools and update schedules. This fragmentation makes it difficult to:

  • Get a unified view of current supply-chain health
  • Identify emerging stress signals before they cascade
  • Share consistent metrics across teams

Approach

Architecture

Built a three-tier system:

  1. Data Layer: PostgreSQL database with SQLAlchemy ORM and Alembic migrations for schema versioning
  2. API Layer: FastAPI backend exposing RESTful endpoints for data ingestion and retrieval
  3. Visualization Layer: Streamlit dashboard with interactive charts and filtering

Data Pipeline

The backend automatically:

  • Ingests data from multiple indicator sources on configurable schedules
  • Normalizes and validates incoming data against defined schemas
  • Computes derived stress metrics (rolling averages, z-scores, trend indicators)
  • Stores historical snapshots for time-series analysis

Dashboard Features

The Streamlit frontend provides:

  • Real-time indicator cards showing current stress levels
  • Interactive time-series charts with adjustable date ranges
  • Drill-down views by region, commodity, or transport mode
  • Export functionality for reports and further analysis

Stack

| Layer | Technology | |-------|------------| | Backend | FastAPI, Python | | Database | PostgreSQL, SQLAlchemy, Alembic | | Frontend | Streamlit, Plotly | | Data Processing | Pandas, NumPy |

Results

  • Consolidated 5+ data sources into a single unified dashboard
  • Reduced time-to-insight from hours of manual aggregation to real-time updates
  • Enabled consistent metric definitions across analysis workflows

Key Learnings

  1. Schema migrations are essential — Alembic saved significant headaches when indicator definitions evolved
  2. API-first design pays off — Decoupling data ingestion from visualization made both easier to iterate on
  3. Streamlit is powerful for internal tools — Rapid prototyping without frontend complexity, but requires careful state management