← Back

Work

AI capability frameworks, measurement methods and hands-on technical work.

Featured work

More projects

Parallel Research: Multi-LLM Research Inside Claude Code

A Claude Code skill that sends one research question to four AI models at the same time and saves their answers into the project you're working in.

AI Systems

writing-style

An open-source Claude Code skill that strips the AI tells out of written output, with a committed eval set so the claims are checkable.

Open Source

Startup-Analyser

An open-source fork of a private idea-evaluation pipeline: capture, score and usually kill a venture idea properly.

Open Source

PokeGrade

A trading-card grading tool that measures what can be measured with computer vision, hands the rest to a model and refuses to bluff a top grade from a single photo.

AI Systems

Claude Deck

A native Stream Deck plugin that turns a bank of physical keys into a status light for each parallel Claude Code session: running, waiting or done.

Tooling

Interactive DCF Simulator

A static DCF spreadsheet rebuilt as an interactive dashboard, with scenario toggles and the discount rate as the headline lever, plus a live demo you can use.

Finance

Flow Trainer

A trainer for learning unfamiliar query languages through progressively harder challenges, built to practise comprehension under pressure rather than memorised syntax.

Tooling

Pitch Dojo

A two-minute pitch warm-up: get grilled by a partner on the questions that decide a raise, then flip seats and learn to read a partner's answer back.

Tooling

Customer Health Console

An operations console where a Customer Success operator asks about the portfolio in plain English and an AI agent works through the data, highlights the at-risk accounts and drafts a recovery plan for the operator to approve.

AI Systems

Campaign Brief Builder

A marketing-operations tool that drafts a full campaign brief from one line of intent (objective, audience, channel mix and localised hero copy for each market), then hands it to a person to edit, check and approve before anything compiles.

AI Systems