Skip to main content

Selected work / Software engineering

The systems. The trade-offs.

Eight software engineering case studies spanning AI developer tools, MCP servers, AI workflow automation, search visibility infrastructure and data pipelines. Each explains what was built, what was tested and what remains uncertain.

01 / Selected work

A closer look at the decisions.

Start with an automotive research prototype, then explore the systems by domain and maturity.

02 / Project index

More work, different constraints.

  1. AI search & SEO infrastructure

    DigestSEO MCP Suite

    Four open-source MCP systems behind DigestSEO: AI visibility, Google Search Console, technical web validation, and live news and search-trend intelligence.

    In production
  2. Developer tooling

    Nofax

    Open-source local human-approval workflows for AI agents with a deliberately narrower remote Worker for one-way notifications and safe request inspection.

    Open source
  3. Developer tooling

    GodPrompt

    Open-source AI coding workflow and companion MCP server: nine task types, a six-stage engineering method, progressive disclosure, and explicit verification gates.

    In production
  4. Operations

    Flight Deck

    Project overview for an email-triage design covering classification, routing, draft preparation, and human-review paths without publishing unsupported outcome metrics.

    Project overview
  5. Publishing

    Content Engine

    A daily publishing pipeline — ingest, draft with an LLM, route through human review, publish and distribute.

    In production
  6. Market data

    Agentic Market-Data Pipeline

    Scheduled, unattended market-data extraction with hard freshness gates and grounded LLM reading, through to auto-published commentary.

    In production
  7. Market data

    Walk-Forward Crypto System

    A public paper-trading experiment with every number traceable and seven failed ideas documented rather than deleted.

    Public experiment

03 / What comes next

Have a system worth getting right?

I design and build connected software, AI tools and automation with explicit boundaries, testable behaviour and a useful handover.