Case study
Migrating prediction services onto the delivery pipeline
A case study about supervising the migration of prediction services — such as a pricing forecast backend — onto a shared delivery pipeline, and making them robust.
All writing
Case studies come first. They come out of real platform work — retail measurement at NielsenIQ, motor insurance at Allianz Technology — and they name the trade-offs instead of skipping them: what the system was, what it cost, and what I would do differently.
Posts are shorter. Tooling choices, infrastructure notes, and the ordinary work of owning something you have to maintain.
Everything published is listed here, grouped by type, and each group links to its own archive so you can read one kind straight through.
How problems from real platform work actually got solved: binary data delivery for retail measurement, migrating prediction services onto a shared pipeline, and the foundations AI initiatives depend on.
Case study
A case study about supervising the migration of prediction services — such as a pricing forecast backend — onto a shared delivery pipeline, and making them robust.
Case study
A case study about engineering the data and delivery foundations that AI-driven initiatives depend on — reusing the pipeline, not rebuilding bespoke paths.
Case study
A case study about querying retail measurement data out of databases and delivering it as binary payloads through a delivery pipeline — filters, enrichments, and file handling at scale.
Shorter notes on tooling, infrastructure, and the ordinary work of shipping software you own — including why this site is static and how it is built.
Post
Notes on moving this site’s diagrams to render at build time instead of in the browser, and on the three classes of bug that only showed up once something was measured properly.
Post
A practical note on choosing GitHub Pages, GitHub Actions, and Hugo for a personal engineering site that stays inexpensive to run and easy to own.