Product, architecture and scaling decisions are cheap to get right at the start and expensive to unwind a year in. I made them under load — a platform at a billion events a day, a team of fifteen, an acquisition — and now I sit in the room when a startup has to make one.
Everything below is the same work read two ways: one column is the business, the other is the technology. Most calls need both.
Senior product & technology advice at a startup-friendly 175 PLN/hour.
Developer, then CPO, then CTO of the company I co-founded. The advice on scaling, architecture and product decisions comes from building and running the thing — not from consulting alone.
When you need CTO-level thinking, not a full-time hire. I take the decisions you would hire a CTO for — architecture, hiring and the technical bar, build-versus-buy, vendor and cloud choices, the technical side of a raise or an acquisition — one to four days a month.
Fractional CTO pageWhen the architecture, the scaling or the technical debt needs an outside read. Two to three weeks: what breaks first at 10×, what the debt costs you in weeks per month, which risks are real, and a prioritised plan your engineers can start on the following Monday.
How an engagement runsWhen the roadmap, the pricing or the priorities need a decision rather than another workshop. What to build, what to kill, what it costs and who it is for — prioritised against revenue, with the trade-offs written down.
Product consulting pageStrategy, pricing, market and the product organisation.
Technology track ↓Systems, delivery, scale and the engineers who run them.
Business track ↑Where startups call me in
Everything else I help with
Track record
Research & development — WDFT, co-owned
Card fees make small payments uneconomical: roughly 0.25 PLN plus 1–3% per transaction, so nothing below a few PLN is worth selling online. This standard takes no percentage at all — a flat fee per package of transactions, designed at 5 USD per 1,000, which brings the cost of a payment close to that of an Elixir transfer.
An open standard that uses the banks' existing rails and mobile apps — deep links and QR codes — instead of building another payment network. Four parts: a registry holding the bank directory, the webhook verification keys, an API and the mobile-app manifests; an SDK that generates payment links and verifies signatures; banks and providers that execute and notify; the implementor that creates the link and receives the webhooks.
sdk · flutter · dart · kotlin · swift · typescript
Instead of one subscription or nothing, a reader commits around 0.50 PLN per article inside a session and settles it with a single transfer later — which monetises the readers who would never subscribe.
Short-lived sessions with small commitments, aggregated into a single settlement and paid out to partners. Four roles — user, operator, partner, bank — and a settlement step that has to stay idempotent.
status · runs in production as GetViaMsg
What I run today
AI, two ways
Most teams buy AI and then wonder where the value went. I help pick the two or three workflows where it pays for itself — and where it is a toy.
Two of my own products are AI products, so this is operations, not theory.
jq instead of a database. The oldest idea on this page, and still the one I reach for first.
What to do next
A task board where everyone gets one vote, the top task is the one you pick up, and only its author can close it. Built because prioritisation arguments are boring.
A widget that orders a callback
A phone number in, a call back out: the whole backend lived with the telecom operator and I built the front. Made for Telecube as a barter deal, as a deliberate answer to CallPage.
A dozen more, mostly experiments
Payment and SMS libraries from a decade of integrations work, and small tools built for one afternoon's curiosity. Nothing that needed a paragraph of its own.
Tell me what you are building, what it costs you today and what happens if it goes wrong. A thirty-minute call, then I come back with what I would do first, in writing. If I am not the right person, I will say so in the first reply.