Data Engineer - Oracle to PostgreSQL Re-Platform ( 102-08SENG-02 )

This is an App DBA role within a re-platforming initiative moving a production, regulated healthcare SaaS platform off Oracle and onto Amazon RDS for PostgreSQL. Application behaviour and the user interface must be preserved so that the platform's existing certification is not disturbed. The estate is multi-tenant and multi-terabyte, with a large in-database procedural body and high daily change volume.

Correctness is engineered in, not inspected in: every converted package must pass an automated dual-execution equivalence comparison against both Oracle and PostgreSQL before acceptance, and every tenant must pass reconciliation before Oracle is decommissioned for that tenant.

Delivery follows an AI-Directed model. Engineers author the specifications and the definition of equivalence, direct AI agents through the high-volume conversion, and hold every acceptance gate — no change is accepted without human validation.

What you will do

• Run and remediate schema and stored-procedure conversion with AWS DMS Schema Conversion (successor to AWS SCT), working the emitted action codes to resolve constructs needing manual redesign.

• Convert PL/SQL to PL/pgSQL at volume under agentic orchestration, authoring the specs and reviewing every output.

• Redesign Oracle-specific constructs with no PostgreSQL equivalent — autonomous transactions, dynamic SQL, external file and network calls, multi-cursor returns — validated against captured behaviour fixtures.

• Build and operate the dual-execution equivalence harness, with agreed tolerances for money, floating point, timestamps, NULL handling, and ordering; wire a both-engine regression suite into the delivery pipeline as an acceptance gate.

• Configure AWS DMS full load plus CDC, stand up the target topology (Multi-AZ, reporting and cross-region read replicas), and execute blue/green per-cohort tenant cutovers with reconciliation gates and soak.

• Deploy client-specific database objects additively per tenant, never overwriting a tenant with a consolidated baseline.

• Rebuild the data-access layer on Npgsql: apply the package-to-function rename mapping across call sites, re-parameterize concatenated SQL, and close the Oracle type-enum leak behind a neutral boundary.

• Convert report procedures to PL/pgSQL and validate regulated reports field-by-field against golden masters; migrate secrets and connection strings to AWS Secrets Manager and KMS.

• Produce the runbooks, mappings, and reconciliation evidence that are contracted deliverables, and support knowledge transfer.

Required

• 5+ years in relational data systems, with verified hands-on experience in both Oracle and PostgreSQL — not necessarily on the same migration project, but real, verifiable production depth in each engine.

• Deep hands-on Oracle PL/SQL: packages, REF CURSORs, autonomous transactions, dynamic SQL, external calls, and the performance behaviour of large procedural bodies.

• Strong PostgreSQL and PL/pgSQL, including the practical semantic gaps versus Oracle (transaction control, NULL and empty-string handling, numeric and timestamp behaviour, implicit casts, ordering guarantees).

• Experience with large-scale, complex database migration or modernization projects — the migration itself is what matters; it does not need to have been Oracle-to-PostgreSQL specifically.

• Working knowledge of AWS DMS and/or Schema Conversion Tool (SCT) — this specific toolset is the core requirement. Broader AWS platform experience (RDS, EC2, IAM, etc.) is preferred but not a gate.

• Python or comparable for comparison harnesses, reconciliation tooling, and test generation; Git and CI/CD.

• Rigour with data correctness — designing tolerances, reconciling across engines, defending an equivalence claim with evidence.

• Written English strong enough to author specifications and acceptance evidence for a regulated client.