Amru – PostgreSQL, Azure Functions, SQL

A US university was expanding its data and AI platforms and needed LLM usage to stay accurate, controlled, and within budget across teams. Amru led AI-assisted engineering for these platforms and owned architecture, code review, and validation. He built LLM usage metering with budget circuit breakers that block over-budget requests before they reach the provider. He also set shared standards so teams could reuse tested cost and telemetry controls, and he delivered a grants workflow on Azure with case authorization, sponsor-approved budget rules, and human approval steps. His PostgreSQL/pgvector retrieval tuning raised the top-five hit rate from 48% to 74% and cut query time from 968 ms to about 245 ms.

• Raised PostgreSQL/pgvector retrieval top-five hit rate from 48% to 74% and cut query time from 968 ms to about 245 ms on a 27-question benchmark.
• Built configurable LLM budget circuit breakers that block over-budget requests before dispatch, linking provider usage to fleet, project, and model budgets for CIO cost oversight.
• Set organization-wide AI engineering standards for usage accounting, pricing consistency, budget enforcement, and telemetry.
• Added cross-repository contract checks so teams could reuse tested controls instead of rebuilding them.
• Led architecture, code review, and validation of a grants platform built on Azure Functions, Microsoft Foundry, and Ellucian Experience.
• Implemented case authorization, sponsor-approved budget rules, and human approval controls in the grants workflow.

A US university was moving its data integrations and reporting off SQL Server and SSIS and onto Azure Functions and a PostgreSQL-based Ellucian Banner Cloud. These systems feed student services, billing, reconciliation, early intervention, and institutional reporting, and the legacy jobs ran into overlapping runs and HTTP timeouts that risked duplicate student charges. Amru led the move: he migrated 20 pipelines and modernized 40 reporting procedures, rebuilding billing and reconciliation with safeguards against double charging and processing that can recover after a failure. He also extended early-alert data feeds with idempotent ingestion and privacy controls, and prepared costed data warehouse proposals for the institutional research team.

• Migrated 20 SSIS/SQL Server pipelines to Azure Functions and led modernization of 40 reporting procedures for PostgreSQL in Ellucian Banner Cloud.
• Added safeguards against duplicate student charges and recoverable queued processing for meal-plan billing and Florida Prepaid reconciliation, addressing overlapping runs and HTTP timeouts.
• Expanded Early Alert integrations to include appointments, event attendance, and housing records through idempotent ingestion and privacy controls.
• Authored costed Azure PostgreSQL warehouse proposals for Institutional Research, defining validation gates and fallback options for network and authentication dependencies.

A multi-entity business group was consolidating its business and accounting records onto a single Spectrum ERP system. That meant pulling data from five entities, each holding about 100,000 records, whose transaction schemas didn’t match across job costing, billing, payroll, banking, and subledgers. Amru owned the data side of the move: mapping, transformation, and validation from discovery through cutover. He also built a UAT sandbox and reconciliation rules so the accounting balanced after migration across all five entities.

• Owned data mapping, transformation, and validation for a Spectrum ERP migration across five entities with about 100K records each.
• Kept data intact from discovery through cutover.
• Built a UAT sandbox for testing the migration before go-live.
• Defined accounting reconciliation rules for debits, credits, job costing, billing, and payroll.
• Reconciled bank balances against subledgers across incompatible transaction schemas.

A global bank needed reliable financial data for regulatory reporting and AML/KYC compliance across operations in several countries, and noisy data was generating false-positive alerts. Over six years Amru worked his way up from accounts payable systems analyst to Lead Data Quality Analyst (VP). He managed source-to-report lineage, ETL mapping, SQL validation, and data-quality rules for priority federal reports. He designed SQL and Ab Initio data-quality controls that cut false-positive regulatory-reporting alerts by about 15%, and he led testing for 12+ quarterly releases with zero compliance-engine downtime.

• Led AML and KYC data-clearance work across U.S., Poland, India, and Philippines operations, applying data-residency controls to cross-border data use.
• Managed lineage and ETL mapping for priority federal regulatory reports.
• Coordinated source-level remediation and SQL validation to support data-clearance decisions.
• Designed SQL and Ab Initio data-quality controls that reduced false-positive regulatory-reporting alerts by approximately 15%.
• Led UAT and PVT for 12+ quarterly releases with zero compliance-engine downtime.
• Automated financial reporting and reconciliation in Oracle iSupplier and accounts payable, applying accounting controls to enterprise workflows.