Python
Local Database
Local ELT Pipeline
Power BIThe Problem
The contact center had no consistent way to see how it was performing against its Service-Level Agreement. Call data existed, but it was raw, disconnected, and required manual handling to turn into anything leadership could review. There was no cloud data platform in place to lean on — the entire pipeline had to be built from the ground up, running locally.
How the Pipeline Works
Key distinction: Unlike the CFPB dashboard, which pulls automatically from a cloud data platform via API, this pipeline runs entirely locally, end to end. That trade-off meant taking on work a managed platform would normally handle — local scheduling, local storage, and hand-built transformation logic — to get the same reliability.
Extract
Raw call logs and interaction records are pulled from the source system and staged for processing — no cloud connector to lean on, so extraction logic is hand-built and scheduled locally.
Transform
Custom Python scripts clean and standardize the raw records, then calculate the core SLA metrics: handle time, wait time, and whether each call met its target.
Load
Processed data is loaded into a locally managed database, giving the pipeline a structured, queryable home without relying on cloud infrastructure.
Model & Publish
Power BI connects to the local database, applying the data model that powers the SLA dashboard — turning thousands of raw call records into a handful of decision-ready metrics.
What This Demonstrates
- End-to-End Pipeline Ownership: Designing and building an ELT process from scratch, without a managed cloud platform handling ingestion, storage, or scheduling.
- Python Engineering: Writing the extraction and transformation logic by hand — the work a cloud connector would otherwise automate.
- Database Management: Standing up and maintaining a local database as the pipeline's structured data layer.
- Business Impact Focus: Connecting the technical build directly to a measurable outcome — $650K+ in annual workflow savings from replacing manual reporting.
Key Findings
- SLA Performance is Strong, but There's a Gap: 92% of all calls met the Service-Level Agreement target in 2021, leaving roughly 4,100 calls — 8% — as missed commitments worth tracking closely.
- Wait Times Aren't the Bottleneck — Handling Time Is: Customers wait only 17 seconds on average before reaching an agent, but the call itself takes nearly 5 minutes to resolve.
- Longer Calls Drive SLA Misses: Days with longer average handle times consistently show lower SLA compliance, pointing to handle-time reduction as the fastest lever for improving performance.