Piotr Jodko — data architect, on contract

The honest version

Most of these projects fail. I've been on both kinds.

The industry has quoted an eighty-percent failure rate for as long as I've been in it, and my own record roughly agrees. I can't change the odds just by turning up; nobody honest can promise that. What I can bring is the pattern recognition of someone who has watched both kinds of project from the inside, and a preference for the gloomy estimate that holds over the cheerful one that doesn't.

Engagements

Three ways to engage.

Fixed scope where scope can honestly be fixed, realistic dates everywhere else. Paid as we go — the day you stop needing me is the day you stop paying.

Delivery audit

For a platform, warehouse or migration that is late, unclear, or making people nervous. I read the code and the backlog, talk to the people building it and the people waiting for it, and dig the real timeline out of the history. You get a written assessment and the handful of changes that would matter most. Occasionally the finding is that you don't need me.

2 weeks · fixed fee
fee TBD

Architect for a season

Three or four months embedded in your team, around thirty hours a week. Enterprise data model, data mesh or lakehouse design on Databricks and Azure, lineage and governance that people actually use, frameworks that generate the boring code so nobody has to type it. I work in the codebase, not in a deck about it.

3–4 months · ~30 h/week
day rate TBD

Rescue

A delivery that has stalled and needs someone to take responsibility for it until it's in production and your own team can run it without me. Scoped after a short audit. It starts hands-on and ends with me leaving — that's the point of it, not a side effect.

scoped after an audit
milestone-based

How I work

Things I ask about before saying yes.

Less a list of demands than a description of where I've been useful. Engagements that go well tend to share these, so I'd rather ask early than discover late.

Something tangible reaches production. Advisory-only work isn't where I'm at my best.
Version control for everything, automated deployment, small releases — DORA, more or less.
A real IDE, Git, and AI-assisted coding where it helps.
A Mac-friendly environment.
Agile as a practice rather than a ceremony; documentation in Markdown, not .docx.
Platform decisions made by the engineers who will live with them.
A budget or rate range stated before the first call, so neither of us wastes an afternoon.
No permanent position, and no managing people as a job.

I have managed people — fourteen months as Head of EU Data Engineering at a large bank, five hundred-odd people going through an Agile adoption. It taught me a great deal, mostly about which half of leadership I'm suited to. I lead by doing the work and then handing it over.

Evidence

Where this comes from.

Most of these were through consultancies rather than direct employment, which is normal in this trade. The end clients are named on the CV.

2026 –Cloud Drift — data platforms for the consultancy's clients, part-time.
2025 – 2026EcoVadis (via Pragmile) — Enterprise Data Architect. Enterprise data model, data mesh on Databricks.
2021 – 2025Objectivity, later Accenture (via ITFS) — Data Solutions Architect. Azure Synapse, Data Factory, Spark, dbt, Python.
2020 – 2021HSBC — Head of EU Data Engineering. See the note above.
2019 – 2020Tracr, De Beers' diamond-traceability platform (via Gravity9). FastAPI, MongoDB, Kubernetes.
2013 – 2018Compact Solutions — metadata, lineage and governance for banks and insurers in the US, Canada, South Africa and Singapore.
2010 – 2013Sabre — Lead Data Architect.
1998 – 2010Asseco / Softbank — developer to architect. The ELT framework behind a large Polish bank's warehouse, and the SAS framework in the first essay below.

Built and shipped on my own time: Blinkomat / BlinkD (ATM access for visually impaired users, 2021), fingerAi, and NidraLab, a yoga-nidra app that is still in progress. Testimonials will appear here when clients write them, not before.

Writing