Case studies

Public-safe evidence, not confidential detail.

These cases describe decisions, methods, and verified outcomes while preserving employer, client, and system confidentiality.

Applied AI research

Evaluating binary and ternary retrieval under a smaller memory envelope

A deployment-fit study of low-bit retrieval for constrained infrastructure — and a case study in catching an overclaim before it shipped.

Agentic AI governance

An operating model for agentic workflows that emphasizes authorization, verification, reversibility, and evidence.

Enterprise product operations

A product-operations approach to reducing risk and cost across a complex application estate.