Adrian Vance Writes

Adrian Vance writes about applied AI, model evaluation, and decision systems for technical audiences. Background in operations research: about a decade building optimization models, allocation frameworks, and decision support tooling for large organizations, in Python, R, and SQL.

The recurring subject is the gap between a model that is technically correct and a system a decision maker will actually use.

---

Writing

1. You Don't Have a Ground Truth Set. Evaluate Anyway.
   Mining decisions humans already made, instead of authoring an answer key.
2. The Objective You Optimized Is Not the One You Wanted
   Multi-objective tradeoffs, efficient frontiers, and why single-objective optimizers lose the room.
3. Stateless by Design: What "Give the Model Memory" Actually Means
   What "give the model memory" actually resolves to, and why fine-tuning is the wrong tool.

---

Available for technical blog posts, long form explainers, and whitepapers with data, analytics, and AI infrastructure teams. No listicles.

adrian@adrianvancewrites.com