Aland Baban

A full-stack restaurant ordering and operations platform: Next.js frontend, ASP.NET Core API, PostgreSQL, SignalR live kitchen updates, role-based dashboards, Docker/DevContainer setup, CI and documented architecture.

Next.js App Router -> ASP.NET Core API -> SignalR -> PostgreSQL -> Docker + CI

ROLE

Product / Full-Stack Engineering

STACK

Next.js / React / ASP.NET Core / EF Core / SignalR / PostgreSQL / Docker

A reproducible AI-engineering reference system built across eight maturity stages: risk/product contracts, retrieval, evaluation, provenance, approvals, observability, security and progressive delivery without provider lock-in.

Risk Contract -> Retrieval + Provenance -> Evaluation Gates -> Approval Workflow -> Observability -> Progressive Delivery

ROLE

AI / Agent Engineering

STACK

LLM Systems / RAG / Agents / Evaluation / Observability / Progressive Delivery

A production-grade, modular auth backend: OTP two-factor login, granular RBAC, token rotation and waitlist management — built as a self-hostable foundation for SaaS platforms instead of another vendor-locked BaaS dependency.

Reverse Proxy -> Auth Service (OTP · RBAC) -> PostgreSQL -> Redis

ROLE

Backend Architecture

STACK

TypeScript / Node.js / JWT / RBAC / OTP / Redis

Become an AI Engineer — Der praxisorientierte Leitfaden für KI-Engineering

Fachbuch für Softwareentwickler, die aus probabilistischen Modellen verantwortbare Produktionssysteme bauen. Vom Product/Risk-Contract bis zu Gateway, Observability, Security und Progressive Delivery — mit dem reproduzierbaren Companion-System SupportPilot (V0–V7, deterministische Fixtures, keine API-Keys).

Graph Engineering: Building AI Agent Systems That Scale Without Breaking

Die meisten KI-Agenten werden noch immer wie Checklisten gebaut. Graph Engineering betrachtet ein agentisches System nicht als einen einzelnen intelligenten Loop, sondern als ein ausführbares System aus Zuständen, Berechnungen und Kontrollflüssen.

A Coordination Layer Architecture for Multi-Agent AI Systems

This paper proposes Orion-Membrane: a zero-trust, selectively permeable coordination layer designed to provide a structured, deterministic, and secure boundary for decentralized multi-agent collaboration.