Active experiments

Current research probes

Open questions in machine intelligence, automated synthesis, and data architecture, tested in working prototypes.

Machine intelligence

Probing model behavior

Testing how AI systems respond to real prompts, edge cases, and constraints, and recording where their outputs hold up and where they break down.

Automated synthesis

Generating working drafts

Exploring how automated tools can turn raw notes into structured summaries, and where human judgment must step in to keep meaning intact.

Data architecture

Mapping information flow

Prototyping how data moves between sources, models, and outputs, with notes on the bottlenecks, dependencies, and failure points I observe.

Applied projects

Working prototypes and system builds

Case study

A prototype under real use

I built a small tool to test how an AI assistant handles structured research notes. The working notes record which tasks it completed reliably, which required correction, and the constraints that shaped each result.

System build

Documenting the architecture

A data pipeline built to move information between a source archive and an output summary. The build log shares both what functioned as intended and the observed constraints that forced redesigns along the way.