Ruben Haisma

I build production-grade AI systems that go beyond prototypes and deliver demonstrable business value. My work sits at the intersection of LLMs, AI agents, and scalable software architecture, translating advanced AI capabilities into reliable systems that run within real organizations — not just in demos.

Over the past few years, I've designed and deployed end-to-end AI solutions in startup and research environments — from self-supervised vision models for medical imaging to RAG-driven AI agents connected to business knowledge, databases, and internal tools. I work full-stack, combining model development, backend systems, cloud infrastructure, and user-facing applications into a cohesive architecture.

My expertise lies in RAG pipelines, LLM integrations, AI agents, vector search, and MLOps, combined with cloud deployment and production backend engineering. I work with Python, TypeScript, FastAPI, Next.js, and PostgreSQL/pgvector, building AI systems that are scalable, observable, and maintainable in production environments.

My focus is on the engineering side of applied AI: turning models into working systems. This means safely integrating language models into applications, connecting AI to structured and unstructured business data, designing decision logic around model output, and building hybrid systems where machine learning, rules, and software architecture converge. The goal is always the same: AI that remains robust, controllable, and usable in daily operations.

Previously, I worked on Generative AI for structured data modeling, developing pipelines that convert database metadata into technical data models via LLMs, validation layers, and schema-driven output. This has built strong experience in structured generation, model constraints, and translating complex technical input into artifacts that engineering teams can directly work with.