Ahmed Hobeishy

• Built the agent harness's tool calling and human-approval flow, turning camera events into SOP-governed actions. • Built RAG over each customer's SOPs and site context, so every agent decision cites its procedure, in Arabic and English. • Built the evals for agent decisions and VLM answers, run on every release. • Deployed the agent services to the cloud and to on-site Deqa Box edge devices, with tracing.

Multi-tenant legal-tech SaaS: production RAG, product and platform. • Built production RAG on the Claude API over each tenant's legal documents, with structured, validated outputs feeding the product. • Automated the tenant lifecycle across services and ran the Kubernetes/Helm development environment. • Led an accessibility refactor that passed a third-party audit, and delivered features across the React/Angular front ends and Python/Node APIs.

Multi-tenant SaaS on AWS: web, mobile and IoT. • Shipped full-stack features across multi-tenant SaaS products, including an e-commerce platform (React, Angular, Spring Boot, PostgreSQL), and built a cross-platform React Native app for real-time IoT device management. • Deployed and operated production on AWS with CI/CD pipelines, and resolved production outages.

Selected for the Linux Foundation's competitive Open Mainframe Project mentorship, working with IBM and Broadcom engineers.

Thesis: how to structure specifications for AI coding agents, a quality and cost study.

Excellent with Honors