dg↗David GellerData & decisions
← All selected work

Python · Analytics automation

Make the routine
run reliably.

A portfolio of recurring workflows that reduces manual repetition and supports consistent analytical delivery.

The principle

Build for the next run.

01

Automate

02

Validate

03

Monitor

The question

Recurring exports, workbook checks, file preparation, and model scoring can consume time even when each individual task looks small. The goal was to reduce repetition while keeping routine outputs dependable.

The approach

I led an automation portfolio of more than 10 recurring Python processes, most running daily, saving approximately 75 hours annually. I also migrated eight browser-export, data-extract, Excel-validation, data-cleaning, and model-scoring automations from R to Python.

Reliability in the workflow

  • Automated browser navigation, downloads, and data extraction using Selenium and Playwright.
  • Workbook conversion, validation, updates, and backups.
  • Logging, run identifiers, error handling, and failure screenshots to support troubleshooting.
  • Scheduled execution and file preparation for downstream processes managed by the BI team.

What changed

The R-to-Python migrations improved maintainability, logging, error handling, and consistency. These workflows support analytics delivery; they are distinct from ownership of production warehouse architecture or BI-managed ETL pipelines.

What matters

An automation earns its place when its failures are visible and its outputs can be checked.

Keep exploring

Explore MetricForge, a personal BI workbench.

Next case study ↗
Powered by Keystone