Representative work

• PowerShell systems administration toolkit

  A large, categorized library of PowerShell scripts for Windows, Microsoft 365, Azure, and Active Directory administration — covering account lifecycle, mailbox and calendar management, security and NTFS permissions, software inventory, and system maintenance.

  Challenge

  Day-to-day administration across identity, cloud, and Windows systems handled with scattered, inconsistent one-off scripts.

  Approach

  A standards-driven collection where every script ships with comment-based help, parameter validation, -WhatIf safety, and idempotency — validated in CI with PSScriptAnalyzer and GitHub Actions.

  Outcome

  Repeatable, auditable, production-safe automation that a team can run and maintain with confidence.

  PowerShell Microsoft 365 Azure Active Directory PSScriptAnalyzer GitHub Actions

  View public code on GitHub (opens in a new tab) →

• MandelaReport — webpage change tracking with AI summaries

  A self-hosted web app that compares a page's live content against historical Wayback Machine snapshots, highlights word-level differences, and generates a plain-language summary of what changed using a local LLM — with a rule-based fallback when no model is available.

  Challenge

  Detecting and explaining how a web page changed over time without tedious manual comparison.

  Approach

  Robots-aware fetching and Wayback snapshot selection, word-level diffing, and an optional local llama.cpp model that narrates the changes — packaged as a Dockerized HTTP API with SQLite storage.

  Outcome

  A clear, human-readable report of what changed between any two points in time, runnable entirely on your own infrastructure.

  Python Docker SQLite llama.cpp (local LLM) Wayback Machine

  View public code on GitHub (opens in a new tab) →

• Automated market-monitoring services

  Python services that monitor Kraken markets — scanning USD pairs for oversold mean-reversion setups and watching for newly listed trading pairs — and send email alerts when conditions are met.

  Challenge

  Catching time-sensitive market conditions without manually watching feeds around the clock.

  Approach

  Scheduled Python services query the exchange API, evaluate conditions, and send SMTP alerts — isolated in a dedicated virtualenv and run as systemd services in an unprivileged Proxmox LXC, with logs surfaced through journald.

  Outcome

  Hands-off, always-on monitoring with alerts delivered automatically.

  Python 3.11 Kraken API systemd Proxmox LXC SMTP alerts

  View public code on GitHub (opens in a new tab) →

Additional client work is available to discuss during discovery where confidentiality allows.

See more on GitHub (opens in a new tab) →