Francesco ZinghinìParty-appointed technical expertise

Projects

The most honest way to show how I work is to show code anyone can read, run and argue with. Here are two public projects, both in Python and both under the MIT licence.

  • italian-legal-interest — statutory interest and monetary revaluation

    Python · MIT licence · public code

    An implementation of Italian statutory legal interest under article 1284 of the Civil Code together with monetary revaluation on the ISTAT FOI index: the two calculations that, combined, make up the classic Italian «revaluation and interest» computation found in employment claims, overdue debts and restitution ordered by a court.

    The technically interesting part is that the statutory rate is set annually by ministerial decree and has changed many times — it stood at 5% for decades, rose to 10% in the early 1990s, and swung between 0.01% and 5% within two years recently. A period spanning several years therefore has to be split, each segment charged at its own year's rate: the library holds the full historical rate table and does exactly that, pro rata on actual days.

    It is the kind of computation that ends up in pleadings, and it has to be done in a way the other side can check line by line. The code is public precisely so that it can be checked.

    • art. 1284 Civil Code
    • ISTAT FOI index
    • pro rata die
    • historical rate table
    • Python
    • MIT

    The code on GitHub →

  • mining-profitability-formula — proof-of-work profitability model

    Python · MIT licence · public code

    An implementation of the standard profitability model for proof-of-work mining: daily revenue as a function of one's share of network hash rate, energy cost, fees, and the break-even electricity price.

    The difficulty adjustment mechanism is, in systems-theory terms, a feedback loop: the network tracks a constant block time by correcting the target in response to total hash rate. The library exposes that adjustment explicitly, so that scenarios can be modelled rather than a single number returned. It is a methodological choice that holds elsewhere too: a model is useful if it can be interrogated, not if it gives an answer.

    • deterministic model
    • feedback loop
    • scenario analysis
    • Python
    • MIT

    The code on GitHub →

The work that cannot be shown

The bulk of what I have done over twenty years is not publishable: production infrastructure, migrations, architectures designed for clients who have no reason to appear on a website. I do not publish client names and I do not use other people's marks as decoration — those things are done with the permission of whoever they concern, and I have no need of them.

If an instruction requires my experience in a particular area to be verified, ask me: I will provide the references I am able to provide, directly and in writing.

The CV → · GitHub profile →

Do you have a matter under way?

Tell me what happened and what you need to prove. In a first reply I will tell you whether there is a technical route, what data is needed and how long it takes — before any commitment.

Request an assessment