THINKING OUT LOUD
Notes from
the workbench.
A future home for practical lessons, questions, and ideas about AI engineering. I'm more interested in sharing what I learn than pretending to have all the answers.
COMING TO THE NOTEBOOK
Topics I'm exploring.
These are planned topics, not published articles. I'll add links when the writing is ready.
01
PLANNED
↗What image similarity misses: thinking beyond embeddings
How multiple signals can contribute to a more useful understanding of visual evidence.
02
PLANNED
↗Designing retrieval systems that are easy to evaluate
A hands-on exploration of the choices behind practical retrieval-augmented applications.
03
PLANNED
↗From research prototype to dependable AI pipeline
Thoughts on interfaces, failure modes, and making model behavior easier to reason about.
UNTIL THEN
Explore the work.
There's already a detailed case study on how vision-language models and image forensics can complement one another.
Read the case study