DecisionGraph.py
A small local-first library for keeping AI decisions replayable. Useful when someone asks, later, 'why did the system do that?' and hand-waving is not enough.
ls -la ~/projects
Things I built around the awkward middle of AI work: data quality, audit trails, routing, privacy, cost, flaky tests, and all the other stuff that quietly decides whether a system survives real use.
A small local-first library for keeping AI decisions replayable. Useful when someone asks, later, 'why did the system do that?' and hand-waving is not enough.
A data-quality control layer for AI datasets. It keeps schemas, access rules, and pipeline steps boringly explicit before anyone tries to put a model on top.
A self-hosted cost monitor that catches weird spend before the monthly report ruins someone's morning.
An offline privacy gate for data pipelines. It catches sensitive values, redacts them, and packages the result so teams can prove what happened.
A privacy-aware process map built from Slack, Jira, and GitHub metadata. It tries to show where work gets stuck without turning people into surveillance objects.
A telemetry tape for replaying incidents. Less folklore in postmortems, more 'here is what actually happened'.
A versioned lockfile for outbound network access. Boring on purpose: make egress explicit, reviewable, and hard to quietly drift.
A policy and approval gateway for AI agents. Agents can move fast; this gives teams somewhere sane to put the brakes.
Evidence packs as code for EU AI Act-style compliance work. Less spreadsheet archaeology, more versioned proof.
A CI flake detector for GitHub Actions. It helps separate real breakage from tests that just enjoy wasting everyone's time.
A deterministic compiler for Battery Passport artifacts. It turns compliance data into signed, replayable proof objects.
A self-hosted email intake router. It extracts the facts, applies explicit rules, and avoids turning every incoming request into a vibe check.