About me

Who I am

I am a psychologist, with a master’s degree and training in psychoanalysis, and I now work with Brazilian public data. I build tools that deliver that data ready to check: every figure comes with its source, and the way back to it.

That requirement is not technical fussiness. It comes from two directions that met: from research, where a figure without provenance cannot hold an argument; and from the clinic, where I learned that the dangerous thing is not not knowing — it is believing you know.

What I have built

MCP servers — six servers that give AI assistants live access to data from IBGE (Brazilian statistics), the Central Bank of Brazil, the Federal Senate, DataSUS, and the international databases of the ILO and UNESCO. Every response carries provenance; all six score 100% on protocol conformity, with automated checks that block regression.

R packages — five packages for accessing public data, two of them published on CRAN: educabR, for education data from INEP, and healthbR, for public health.

Dashboards (in preparation) — visualisations of epidemiological indicators computed from official data published by the Brazilian Ministry of Health and IBGE.

A citable dataset — the participation layer of the Federal Senate’s e-Cidadania portal, versioned, with a DOI on Zenodo and a per-field provenance envelope. It is the layer no existing legislative data package covered.

Where I come from

Before I wrote code, I wrote about psychoanalysis. My first published article examines how the concept of transference appears in Breuer and Freud’s Studies on Hysteria — and argues that what the later tradition treated as “the whole relationship between patient and analyst” was, in the original text, something far more specific and far stranger.

That sounds distant from downloading DATASUS microdata, and it is not. In both cases the work is the same: read the source carefully, find where it says less than it appears to say, and never let interpretation pass itself off as description.

How I work

Three things I always do, which explain most of my technical choices:

Provenance before convenience. A number that cannot be traced back to its origin does not enter an analysis of mine. That is why every server and every package hands back the source along with the data.

Verification derived from the source. A count written by hand next to the thing it describes rots in silence. I would rather have a test that confronts the claim with the origin — and fails loudly when they diverge.

A noisy error instead of a plausible answer. When something cannot be answered safely, the tool has to say so. Wrong data under a good-looking label is worse than an error, because it looks like an answer.

Outside of work

I read philosophy and history, watch documentaries, and spend time with my family. I think a balanced life makes for better thinking — and I am wary of people with only one thing in their heads.

Contact

Write to sbissoli76@gmail.com, or find me on GitHub and LinkedIn.