Concepts explored · Thought documentation

Applied work

Concepts explored and documented along the way: data and analytics methods, process and operations, portfolio finance, and the tools I build to run them.

Case studies

Business intelligence

Power QueryDAXPower BI

Designing and deploying live portfolio analytics

From a hand-made weekly workbook to a live Power BI report on a financing portfolio. One model, eighteen pages, and the process rules that had to be written from scratch.

Read article9 min read

Financial analytics

ExcelPython

Aiming a budget cut

A 20% reduction lands on a three year investment portfolio. Spreading it evenly is the fastest way to do the most damage. The classify and aggregate method I use instead, and the tool that runs it.

Read article7 min read

Competitive intelligenceDrafting

Claude CodePower AutomatePower Apps

AI competitor mapping & features tracking

An AI-powered competitive intelligence system that maps the end-to-end scientific journey against a cross-competitor feature matrix, surfacing the structural barriers that point to high-value, equity-driven product interventions.

Not published yet

Product & engineeringDrafting

Claude CodeNext.jsSupabase

Three hundred organisations, and the map between them

Designing and deploying a Next.js and Supabase application to hold 300+ organisations, the networks connecting them, and a multi-stage outreach pipeline, in one place a team can actually decide from.

Not published yet

Process engineering

Python

Process analysis

Reconstructing a reporting process from its timestamps, handling missing runs and outliers, then reading actual step timings against target to find where the time was going.

Read article6 min read

Technical walkthroughs

Project controls

Python

Earned Value Management, made legible

A dashboard that turns a project budget into CPI, SPI and estimate at completion, then plays the same programme out under positive, neutral and negative scenarios so the numbers can be argued with.

Read article8 min read

Process engineering

Python

Measuring a process against its target

Walking a business process step by step and comparing what actually happened against the durations it was supposed to take, with the data generated so you can explore the method rather than the case.

Read article7 min read

Forecasting

Python

Time-series forecasting in practice

From Holt-Winters to SARIMA and Prophet. Forecasting business series and connecting back to Earned Value Management.

Read article7 min read

Anomaly detection

Python

Anomaly detection for financial oversight

Spotting unusual transactions and process deviations with Z-score, IQR fences, rolling windows and KNN-distance. A practical view for audit and compliance teams.

Read article9 min read

Unsupervised ML

Python

Finding structure without labels

An overview of unsupervised clustering in Python. Encoding categorical data, scaling, dimensionality reduction (PCA), and choosing between k-means and DBSCAN.

Read article8 min read