VKR

VKR Research publishes working papers, technical memos, and benchmarks on markets, AI, and computational systems.

The public research surface is artifact-first. Each project should leave behind a memo, dataset, benchmark, market map, or reproducible technical tool that another serious reader can inspect.

Current AI Research work is organized around inference economics, verification, automation infrastructure, compute markets, and deployment risk.

Working Paper

The Cost of Useful Tokens

A measurement framework for pricing AI work by accepted task output rather than generated tokens.

Read

Working Paper

Verification Is the Bottleneck

A taxonomy for understanding which AI work can be checked cheaply enough to scale.

Read