01 / TELECOMMUNICATIONS
Enterprise Telecom Reconciliation Engine
An Azure-based telecom reconciliation solution comparing operational and analytical data from Oracle and Snowflake. The system identifies mismatches and manages discrepancy cases through Delta Lake.
GlobalLogic · Client: T-Mobile · Senior Data Engineer · June 2026 – PresentTHE CONTEXT
A clear starting point.
As a Senior Data Engineer at GlobalLogic, working with client T-Mobile from June 2026, I built ingestion, transformation, reconciliation, and case-processing workflows. I collaborated with data engineers, technical leads, and business teams on requirements, validation, UAT, and releases.
THE CHALLENGE
What needs to work.
Source records differed in format, completeness, and order classification. Missing subscriber identifiers required fallback matching, and date comparisons needed consistent treatment. Recurring mismatches had to update existing cases without creating duplicates, while resolved discrepancies needed to close with a retained processing history.
THE APPROACH
Inside the decisions.
Create a consistent comparison boundary
Built Azure Data Factory and Databricks pipelines to ingest Oracle and Snowflake data, then used PySpark and Spark SQL to standardize attributes before reconciliation.
Handle incomplete records explicitly
Implemented matching and fallback logic using multi-table lookups, null handling, historical date comparisons, and order-type standardization. Internal identifiers, date windows, and proprietary rule definitions are omitted from this public account.
Separate discrepancies from their lifecycle
Designed Delta Lake processing to create cases for new discrepancies, update recurring failures, close resolved cases, and retain processed-record history. Reusable notebooks separated ingestion, transformation, rules, exception handling, and testing.
Validate across environments
Performed schema, null, duplicate, referential-integrity, and source-to-target checks during development and UAT. Supported environment-specific configuration, GitLab CI/CD releases, and post-deployment validation.
THE TOOLKIT
OUTCOMES & VALIDATION
Evidence over assumptions.
Delivered automated cross-system validation and discrepancy detection with case creation, updates, resolution, and processing history. Reusable notebooks and validation controls made the workflow easier to maintain and extend.
A reconciliation result needs a history: what was compared, what failed, and whether the discrepancy remains open.