Evidence over hype
Claims should match the work that can be demonstrated. Experiments, client implementations, and production systems are described differently.
Compiling the next screen…
Compiling the next screen…
About Cataluma
Cataluma is a software engineering consultancy led by David Bleeker. The work combines extensive application architecture and product engineering experience with applied AI systems, intelligent automation, and AI-enabled product development.
What Cataluma Is
Cataluma helps organizations turn promising technology into maintainable software. Engagements span architecture, implementation, integration, evaluation, and deployment, with direct senior involvement throughout the work.
Define system boundaries, data flows, integrations, operational constraints, and the decisions that should remain easy to change.
Build the application, workflow, and integration code needed to turn an architectural direction into working software.
Keep testing, evaluation, deployment, observability, and maintainability inside the engineering scope from the beginning.
Principal Consultant
David is a software architect and applied AI engineer with extensive experience designing and delivering enterprise and commercial applications. His background spans application architecture, full-stack implementation, API and data design, cloud delivery, and technical leadership.
His applied AI work builds on that engineering foundation: connecting models to applications, organizational data, tools, review workflows, evaluation, and operational controls. Client implementations, production systems, and experimental engineering are identified clearly according to the available evidence.
Engineering Philosophy
Claims should match the work that can be demonstrated. Experiments, client implementations, and production systems are described differently.
New technology earns its place by improving the product or workflow, not by making the architecture appear more advanced.
Architecture decisions should make their costs, constraints, failure modes, and operational consequences visible.
Reduce uncertainty early, validate the highest-risk assumptions, and expand implementation when the evidence supports it.
Prompts, models, retrieval, and tools need task-specific evaluation and review—not intuition alone.
Use review, approval, and escalation where consequences warrant them instead of assuming every model output can act autonomously.
Technical Scope
Technologies support the architecture and product outcome; they are not the positioning by themselves. Every engagement uses only the parts of this scope that the problem requires.
LLM integration, RAG, vector search, orchestration, human review, evaluation, and observability
TypeScript, Next.js, React, Python, GraphQL, APIs, state, and workflow design
PostgreSQL, cloud deployment, CI/CD, integration architecture, testing, and operational readiness
Work directly with Cataluma’s principal consultant on the next useful technical step.
Discuss your project