A Monte Carlo forecasting engine for rugby tournaments, refreshed daily.
- Python
- Glicko-2
- XGBoost
- SQLite
- Next.js
- TypeScript
- Vercel
Read the case study Visit
Data and software consultancy
Complexity,
Solved.
I build the data and software systems a business runs on — the pipeline that feeds it, the model behind the decision, and the application that puts both in front of your users.
Approach
Systems your team can run and change.
Your business carries real complexity, but complex shouldn't mean complicated. The thing you use every day should still be simple to run and safe to change.
Most of what I'm asked to fix was built as a utility, often quickly, for a situation that has since changed, and then relied on for years afterwards. The work is usually about making the logic explicit enough that somebody other than its author can change it without breaking it.
Services
What I'm usually brought in to do.
- 01
Data pipelines
Data pulled out of the systems it lives in and landed somewhere you can query it, on a schedule, without anyone watching.
- 02
Models and forecasting
Quantitative models that publish their own accuracy, so you know what to trust them for.
- 03
Applications and dashboards
Software and reporting that produce the answer, so nobody spends an afternoon assembling it.
- 04
Automation
The repetitive parts of a recurring process removed, leaving people the parts that need judgement.
- 05
Spreadsheets, done properly
Built to hold up where a spreadsheet is the right answer, and lifted into code where it has stopped being one.
- 06
Technical leadership
Standing in as the technical lead a small team doesn't have: architecture, review, and the pipeline that enforces both.
Practice
Embedded technical leadership.
I work as the technical lead for companies that need one without hiring one: advising the executive on the build-versus-buy call, owning the architecture, and drafting the scope then delivering against it.
In practice that has meant ETL pipelines feeding time-series stores, dashboards, internal admin systems, and the CI that keeps them honest.
Before consulting, quantitative roles across asset management, management consulting, e-commerce and technology — for large listed companies down to micro enterprises.
The name is where this started. A spreadsheet is usually the first place a business writes its real logic down, so it's still where a good number of engagements begin.
Built
Selected projects.
Consulting work belongs to the client, so these are what I can show. Each one is running, published, or both, and each shows the method as well as the result.
tidylearn
ReleasedOne tidyverse-shaped interface over R's machine learning ecosystem.
- R
- tidyverse
- glmnet
- xgboost
- randomForest
- ggplot2
- gt
Ready Before Run()
OngoingA practical guide to the infrastructure data science assumes you already have.
- Quarto
- R
- Python
Visit Source
Writing
Ces's random twalks on tech and stats.
Easy Come, Easy Go: Building Software When Anyone Can
AI means I can build the tools I need myself. It also means anyone can, so what I build is less likely to add much for anyone else. Both come from the same fall in the cost of building.
Too Much of a Good Thing: Testing Why AI “Sounds Like AI”
In May I argued AI sounds like AI because it writes like the average of us. Four pre-registered studies found the tells are ours, with an open model using some far more than any human writers do.
In the Long Run: Which Doomsday Warnings Are Worth Having
Warnings about disaster do the most good when they say how it would happen and what to do about it. What economic history, warning research, forecasting and AI safety know about that.
Contact
Tell me what's not working.
Send me the problem as you'd describe it to a colleague. I'll tell you what I'd do about it before you have to commit to anything.
cesaire@sheetsolved.com
Cape Town, South Africa