Case Study · 2021
iCLOUD & iDSA
Python pipelines and React dashboards for semiconductor design data — sub-second queries across multi-million-record sets.
Context
Heterogeneous semiconductor design datasets — too big for a spreadsheet, too slow for a naive query. Teams needed live error analysis, trends, and cross-project comparisons.
The Problem
Analysis cycles took too long. PDF/CSV exports were manual. Queries didn't hold at a million rows.
Process
Normalize the sources
Python pipelines to ingest, transform, and normalize design datasets from heterogeneous sources.
Make it interactive
React dashboards with charting, filtering, and drill-down for error analysis, trends, and comparisons.
Make SQL keep up
Query and indexing work for sub-second response across multi-million-record datasets.
Decisions
Drill-down before the summary
A KPI with no path to the error row doesn't help an engineer. Every chart had to open.
Export is a deliverable
Automated PDF and CSV — the analysis cycle doesn't stop at the screen.
Outcome
✦ Sub-second queries across multi-million-record datasets
✦ Automated PDF/CSV reports — shorter analysis cycles
Reflection
“A dashboard is not a chart. It's the shortest path from an anomaly to the row that explains it.”