Looking for a job

Ziad Abdalla

@ziadabdalla

Applied AI Engineer at AiCentive who built a multi-tenant RAG pipeline with an 87x speed-up; now building document AI for M&A.

Egypt

Message

At AiCentive GmbH, I designed and built a multi-tenant RAG pipeline that turns PDFs into a NebulaGraph knowledge graph. Model optimization and asynchronous parallel extraction made the pipeline 87 times faster.

I also built the FastAPI REST layer and a LangGraph retrieval state machine, and migrated the system from Neo4j to NebulaGraph for per-tenant graph isolation.

I'm now building ZiadBrief, document AI for mid-market M&A advisory, on EU-only, GDPR-by-design, tenant-isolated infrastructure. It provides source-cited answers for due-diligence data rooms.

At Key Technical International, I built a Firebase and Firestore backend and delivered a role-based admin CMS with a 3D-interactive React SPA. My projects include document processing with Claude API and a local-first job aggregator with semantic CV matching.

Experience

Work history, roles, and key accomplishments

Current

Founder

Current

ZiadBrief

Apr 2026 - Present (6 months)

Building document AI for mid-market M&A advisory on EU-only, GDPR-by-design, tenant-isolated infrastructure, providing source-cited answers for due-diligence data rooms.

DocumentAI RAG GDPR M&A Virtual Data Rooms

AI Engineer

AiCentive GmbH

Feb 2026 - Jul 2026 (5 months)

Designed and built a multi-tenant RAG pipeline turning PDFs into a NebulaGraph knowledge graph, achieving an 87x pipeline speed-up via model optimization and async parallel extraction. Built the FastAPI REST layer and a LangGraph retrieval state machine, and migrated Neo4j to NebulaGraph for per-tenant graph-space isolation.

Lang Chain LangGraph RAG Knowledge Graphs Faiss Sentence Transformers fastAPI Nebula Graph Neo4j IONOS AI Hub LLaMA Mistral Python Docker

Software Engineering Intern

Key Technical International

Aug 2025 - Dec 2025 (4 months)

Built a Firebase/Firestore backend with real-time listeners and auth, and a multi-layer cache reaching ~95% hit rate and sub-10ms loads. Delivered a role-based admin CMS with inline editing, drag-and-drop reordering, Cloudinary media management, and a 3D-interactive React SPA.

Firebase Firestore React Cloudinary CMS Cache JavaScript TypeScript