Analytics Pipeline

Turned multi-tenant post-purchase events into customer-portal metrics, embedded BI dashboards, scheduled reports, and usage billing.

Tech stack: Python · PostgreSQL · MySQL · Elasticsearch · Django · Auth0 · SQL · Azure SQL · Power BI · AWS S3 · Apache Airflow / Amazon MWAA (POC) · BigQuery (POC) · Looker Studio (POC)
Operating context: 30M+ daily events · 50+ franchises · Hourly and daily batch workloads
Role: Software Engineer, Analytics team
Ownership: Reporting recovery, provider-aware billing, business-segment metrics, and the analytics-modernization proof of concept.

Selected ownership

Reporting lifecycle and safe reruns

Hardened scheduled reporting with explicit lifecycle states, failure visibility, performance controls, and an on-demand rerun worker that prevented duplicate delivery through a conditional claim.

Provider-aware usage billing

Integrated Twilio SMS and WhatsApp usage into daily billing details and monthly pass-through reports while preserving provider, tax, tenant, and reporting-period boundaries.

Business and user-type metric dimensions

Extended franchise-level conversational metrics with business and user-type dimensions across event normalization, PostgreSQL aggregation keys, Azure sync, and Power BI views while preserving higher-level rollups.

Analytics modernization proof of concept

Prototyped managed Airflow orchestration and designed a tenant-aware Looker Studio path around the customer-portal identity boundary before changing production.

The conceptual map explains the production workflow and a clearly labeled proof of concept. Highlighted areas and the records below describe direct work. Other components provide system context, and scale refers to the production environment rather than components authored by one engineer.