data-minimization

WORK ITEM: Repo for Data Minimization and selective disclosure

Local-first AI WAF, compliance gate, and cryptographic provenance engine for agentic workflows. Intercepts payloads, slashes token spend via Prose-Tax minimization, and mints tamper-evident audit chains on local silicon.

A simple data minimization and anonymization microservice wrapped around go-minimizer

Data minimization, pseudonymization, and anonymization helpers for Go

Contains all the code used and submitted for the indiviual assignments of the Data Protection Technologies (https://coursecatalogue.uva.nl/xmlpages/page/2023-2024-en/search-course/course/110258) as part of the MSc in Computer Science (Big Data Engineering track).

Source code of IJCB2024 paper "Controllable Privacy in Face Recognition: A Filter-based Approach"

Organizational Cognitive Telemetry (OCT) — a system for capturing sanitized AI interaction telemetry to identify cognitive friction, knowledge gaps, and usage patterns across an organization, without storing sensitive content or individual identities.

A self-hosted program container for cohort-based learning.

A JavaScript client with a graphical user interface for Kalita, a text-to-speech software with a focus on data minimization and user privacy.

privacy.md — a privacy boundary for personal AI agents, enforced before the data leaves your machine. AGENTS.md tells an agent how to work; privacy.md tells it what it may send.

Data Prism is an open-source privacy layer for Java/Spring teams putting LLM agents or MCP clients in front of internal APIs holding customer data. It pseudonymises personal data per privacy scope, refuses anything unclassified, and can keep a hash-chained audit trail.

Large-scale privacy over-collection detector for voice-assistant apps, using LLM embeddings + anomaly detection to flag apps requesting more data than their functionality needs (ICSE 2024)

What happens when privacy law meets skewed data? A reproducible 960-run computational experiment measuring the joint effects of data minimization and class imbalance on ML performance and fairness. PhD dissertation research.

2026 AI Co-Scientist Challenge Korea(AI 연구 동료 경진대회) /Reproducible experiment code for the AI Co-Scientist Challenge 2026.

Chain-agnostic protocol for W3C Verifiable Credentials with proof-of-concept implementations on Ethereum (Solidity) and IOTA (Move). Credentials are signed off-chain as JWTs, hashes anchored on-chain for tamper-proof verification and transparent revocation. Includes CLI, REST API, and web interface.

A Java speech synthesizer and backend server for Kalita, a text-to-speech software with a focus on data minimization and user privacy.

Decide whether an activity event may be retained and produce a minimized record under an explicit consent/retention policy.

Offline proof of concept: coarse categorical incident postcards from 512-channel classical residual frames, with no raw-data sharing.

Open Agent Privacy Protocol & Runtime for pseudonymization, data minimization, scoped re-identification and privacy-safe AI agent access.

Privacy engineering suite — nine Claude Code skills plus a citation-backed regulatory taxonomy of 29 jurisdictional and sectoral records, bridged by a statutory dissolution map. Compliance is a floor; selective disclosure is the ceiling.