# Tiaan van Antwerpen

> Tiaan van Antwerpen is an AI engineer at Spatialedge who builds agentic AI systems for enterprise knowledge work. He studied industrial engineering (BEng) and then data science (MEng, cum laude) at Stellenbosch University, and measures AI the way an industrial engineer measures a production line: lead time, throughput, cost per case and quality.

This site is his portfolio. Every page below has a markdown version, linked here; https://tiaanvanantwerpen.com/llms-full.txt has all of it in one file.

## Profile

- [Tiaan van Antwerpen](https://tiaanvanantwerpen.com/index.md): who he is, how he thinks, experience, projects, writing and education

## Projects

- [Omni Apps](https://tiaanvanantwerpen.com/projects/omni-apps/index.md): A personal project. A platform where the agent writes work as code, storing validated workflow graphs that can be reviewed, edited and re-run regardless of which model fills each step.
- [A wedding-planning PA](https://tiaanvanantwerpen.com/projects/wedding-planning-pa/index.md): A personal assistant for wedding planning that runs from its own inbox. It corresponds with venues, researches options, and keeps a pricing tracker up to date.
- [LLM Routing for Text-to-SQL](https://tiaanvanantwerpen.com/projects/llm-routing-nl2sql/index.md): An LLM routing orchestration framework for NL2SQL agentic workflows. My MEng research, released open source.

## Writing

- [The factory moved into the inbox](https://tiaanvanantwerpen.com/blog/the-factory-moved-into-the-inbox/index.md): Why building agentic AI is an industrial engineering problem, and what a hike in The Goal has to do with it.
- [The model is not the hard part](https://tiaanvanantwerpen.com/blog/model-is-not-the-hard-part/index.md): Notes from forward-deployed enterprise AI work: the model is one component among many, and rarely the one that decides whether a project succeeds.
- [Adaptive LLM routing for effective agentic workflows](https://tiaanvanantwerpen.com/blog/adaptive-llm-routing/index.md): My MEng research: routing a text-to-SQL agent across thirteen models. The RAG router landed on the Pareto front at a fifty-sixth of the cost of Claude Opus 4, and on correctness alone, none of it beat simply picking a good model.

## Optional

- [Field notes](https://tiaanvanantwerpen.com/adventures/index.md): hiking and the outdoors, away from work
- [LinkedIn](https://www.linkedin.com/in/tiaan-van-antwerpen): professional profile
- [GitHub](https://github.com/tiaan720): code