Abdullah Ahmed

Building an AI-powered email-triage and workflow-automation platform for marketing operations.

• Enforced tenant isolation as a runtime constraint across the data-access layer and designed SNS/SQS async processing with fair distribution and partial-batch failure handling.
• Delivered an asynchronous SQS ingestion pipeline with batching, delay-queue reconciliation, concurrency caps, stranded-document recovery, and idempotent race guards.
• Designed an event-driven RAG inference pipeline with SQS orchestration, Bedrock Knowledge Bases, pgvector retrieval, multi-provider LLM routing, and graceful degradation.

• Established tenant isolation as a platform-wide runtime invariant, requiring tenant context for SQL/JPA operations across a 60-person engineering organization.
• Led reporting decomposition from 100+ slow MySQL views to a BigQuery star schema and governed data-access layer, reducing latency by 40% and error rate by 20%.
• Built scalable batch ELT pipelines across MySQL, AWS DMS, S3, Cloud Storage, BigQuery, Lambda, and Cloud Functions with validation, monitoring, and data masking.
• Applied domain-driven design to decompose monolith capabilities and built infrastructure services for configuration, cloud-resource discovery, and automated tenant provisioning.

Enhanced the Staff Management System supporting Apple’s Sales Consultant program across 500+ retail locations.

• Improved page-load time by 30–50% and data-transfer volume by 20–35% through client-side caching, compression, and lazy loading.

Led a four-engineer team building a virtual-dealer web-services platform for mobile-network operators.

Contributed to an anti-money-laundering platform, collaborating across product and technical teams on requirements, defect analysis, and cross-team implementation.