Selected engineering systems

AI systems designed around real constraints.

Four open-source projects exploring how production agents can use less context, coordinate specialist intelligence, retain durable knowledge, and keep regulated data inside the enterprise boundary.

01

TokenOpt

A sufficiency-gated context compiler that transforms noisy transcripts into the smallest evidence set an agent needs for its next decision.

→ 02

AI Engineer Micy

A VS Code-native, multi-agent engineering system that coordinates quality, security, DRY, testing, planning, and Azure DevOps workflows.

→ 03

Project Memory

A local-first knowledge layer combining structural code intelligence, decision history, research, reusable scripts, and hybrid retrieval.

→ 04

Sovereign AI Privacy Gateway

A self-hosted inspection and tokenization boundary that keeps regulated data out of external models while producing encrypted audit evidence.

→