An illustrated broadcast newsroom seen from above: a studio on air, a video wall, rows of desks and people walking between them. Each reporter wears a name tag with a section of this portfolio; the tags jump to that section below.
Brazilian journalist, researcher and instructor based in London, with a proven track record of producing and managing impactful projects related to Brazil.
My reporting has been translated into several languages, and my video reports have drawn over 36 million views across social media.
Stories I wrote sparked a federal investigation into Jeffrey Epstein's sexual enticement network in Brazil, exposed a former president's misuse of public funds, uncovered a bid fraud that saved R$ 15 million from a city's budget and enabled the shutdown of companies involved in illegal activities.
My BBC reporting on Epstein was also cited in the U.S. House Oversight Committee’s final report on the Epstein and Ghislaine Maxwell investigations (2026, p. 101, notes 584–585).
Attended world-class conferences as instructor in Brazil, USA, UK, Portugal, Malaysia, Sweden and others. Trained over 2,000 journalists through online and in-person workshops.
Masters degrees in Data Journalism (Columbia) and Public Administration (FGV‑EAESP). Research-assistant at POLIS‑Cambridge (2024–2026). Former research fellow at Oxford (2021), Columbia (2022–2023) and NED (2023–2024). My research was used by the government to improve transparency policies.
I enjoy testing different technologies to improve my investigative reporting process and to help other journalists. I founded Datafixers, a data innovation lab that received a US$ 100,000 grant from Columbia and Stanford universities and won the prize for best data journalism project in Brazil.
I analyzed all the millions of Epstein files released by the US government and built scrapers that searched for keywords related to Brazilian figures. This way, we were able to reveal the financial dependency of Brazilian models on the billionaire, connections used to lure young women, and the victims who were taken to the US. The BBC series led to a Federal Prosecution investigation in Brazil seeking a network that lured women. Combined, the reporting reached over 10 million views in text and video and resonated across all major Brazilian media outlets.
In November 2026, the U.S. House Committee on Oversight and Government Reform cited my English-language BBC report twice in its final report on the Epstein and Ghislaine Maxwell investigations (p. 101, notes 584–585). The citations appear in the section on Epstein’s international network and support its account of Jean-Luc Brunel’s modelling agency and the alleged use of modelling work to obtain U.S. visas for young women.
The English version of the investigation — Epstein used modelling agent to recruit girls, Brazilian women tell BBC — was published internationally and translated into several languages, including French (Epstein utilisait un agent de mannequins pour recruter des jeunes filles, témoignent des Brésiliennes auprès de la BBC), Polish (Brazylijskie modelki twierdzą, że Epstein rekrutował młode dziewczyny przez agenta modelek), Serbian (Brazilke za BBC: Epstin koristio modnog agenta za regrutaciju devojaka), Chinese (巴西女性向BBC透露，爱泼斯坦曾利用模特儿经纪人招募女孩), and Ukrainian (Як модельні агенції вербували неповнолітніх дівчат для Епштейна).
With Leandro Prazeres, I examined messages extracted by police that described efforts to remove reporting about Daniel Vorcaro and his allies. The messages included a proposal to access a newspaper’s systems; they did not establish that access occurred.
This investigation mapped a network that generates fake AI doctors to engage elderly audiences. The work led to a police investigation and the takedown of several YouTube channels involved in this kind of wrongdoing.
Our investigation reveals that at least nine girls aged 11 to 13 reported inappropriate touching by a military monitor in a public school, raising broader questions about oversight and the civic-military model.
This investigation dismantled a network of over 10,000 people selling bottle parts for beverage counterfeiting and triggered a police investigation.
The thick, patterned skin of the once-at-risk pirarucu fish is in demand for high-end fashion, but the fishermen want more money.
Investigation reveals how a budget amendment supported by Chamber President Hugo Motta funded construction using stones produced by workers in slave‑like conditions.
Exclusive investigation exposes precarious working conditions and abuse faced by Brazilian workers in Ireland’s meat processing industry.
A 2‑month investigation into animal feather smuggling and how companies benefit from exploiting indigenous communities in Brazil.
In the Distrito Federal, Brazil, a significant issue of land grabbing in conservation areas is emerging, intensified by political support. These activities are particularly rampant in the Colônia Agrícola 26 de Setembro, a settlement approximately 26 km from the Palácio do Planalto, housing around 35,000 residents.
Brazilwood is being driven to extinction by an industry not often associated with organized crime: classical music. Tests on a sample of the confiscated wood show it was logged in a protected forest.
Browse the reporting archive →
From earlier newsrooms — select one to see additional stories not featured above.
In 2026, investigating the impact of artificial intelligence on society has become a central focus of my reporting. These stories look at how synthetic media, automation and AI-generated authority are changing what people watch, how they work, whom they trust and how political influence spreads.
With Rowan Ings, I mapped 37 Instagram and TikTok profiles using synthetic women to promote presidential candidates. The accounts had nearly seven million followers; some directed users to paid adult content. TikTok said it removed content shared by the BBC.
I investigated how synthetic videos turn the history of slavery and racial violence in Brazil into a high-engagement spectacle — and the platforms, incentives and audiences that allow this material to spread at scale.
This investigation mapped a global network of AI-generated doctors using fear to go viral among older people in Brazil.
After the first report, a doctor recognised that his face was being used in one of the channels mentioned by the BBC and decided to file a police report. Police then opened an investigation and cited the BBC report as a source.
In October, Google told Brazil’s government it had removed 46 channels and 11 videos featuring fake AI doctors after a government notification that followed our reporting. Read the October follow-up. Versions of the report were published in several languages: Pashto (Afghanistan), Hindi (India), Thai (Thailand), Russian (Russia), Sinhala (Sri Lanka), Urdu (Pakistan), Tamil (India and Sri Lanka) and Ukrainian (Ukraine).
BBC tests confirmed that Meta AI rejected some image requests using Lula’s name while accepting requests using Flávio’s. With Leandro Prazeres, I reported Meta’s response and the differences across its apps.
How the promise of an AI-powered creator economy is changing work, risk and expectations of financial independence.
An investigation into repetitive synthetic content built for attention — and experts’ concerns about effects that may last a lifetime.
Behind the election: synthetic political fandom, automated WhatsApp bots and the infrastructure of influence.
What happened when chatbots were asked to take a political position, despite rules against doing so.
Following a synthetic persona at the intersection of politics, virality and persuasion.
The hidden use of artificial intelligence in mental-health care and the risks for patients.
‘People are lost’: a story about making an emerging technology legible to a confused public.
Efficiency may improve — but what happens to the human experience of being assessed?
Reporting on Brazil from another country is a challenge. I have increasingly turned to open-source tools and public records to investigate public officials, companies and policy decisions. These stories show what careful verification of digital and documentary evidence can reveal.
I used Meta’s Ad Library to identify paid posts questioning Brazil’s electronic voting system and examine who was paying to distribute them. The reporting found ads linked to candidates and electoral campaign services; some ads were removed after the BBC contacted Meta.
Eduarda Campopiano, elected São Paulo’s most-voted state deputy in 2026, had submitted bills as a city councillor that repeated texts from other legislatures. I compared her 20 original proposals and found matching passages in at least 14, including references to the wrong city and a link mentioning ChatGPT.
With Leandro Prazeres, I examined voting-machine records and compared them with polling-station result bulletins to assess claims about the extension of voting hours. The BBC analysis and the electoral court’s figures put those votes at less than 0.03% of the total.
I rebuilt a website taken offline by a Brazilian court using code its developers had left on GitHub. Checking the page against the case file showed that the election petition had misidentified the conversation displayed: it was not an exchange between Senator Jaques Wagner and banker Daniel Vorcaro.
Read the investigationWe checked a number listed in records from banker Daniel Vorcaro’s phone against a parliamentary inquiry and accounts on TikTok and other apps. The independent matches added evidence linking the number to Justice Alexandre de Moraes, while the police report stopped short of confirming who used the line.
By examining the source code of e-commerce pages, including Hotmart and Kiwify storefronts, I traced people behind channels spreading misleading health advice through AI-generated doctors. The reporting followed how attention was turned into sales of digital products.
Read the investigationBrazilian property records established that Jeffrey Epstein had bought an apartment in São Paulo in 2003 and sold it two years later. The deed provided documentary evidence of a previously little-known connection to the country.
Read the investigationOpen-source research helped uncover a case that had not previously been reported: at least nine girls, aged 11 to 13, alleged that a military monitor had touched them inappropriately at a school in Paraná. The court case remained under seal; the monitor had not been convicted when the story was published.
Read the investigation
BBC News Brasil · 23 October 2026
BBC News Brasil · 8 October 2026
BBC News Brasil · 1 October 2026
⚠️🥃 Como funciona a venda de lacres falsos, tampas e garrafas de bebidas alcoólicas no Facebook
⚠️🤖 Como não deixar o ChatGPT atrapalhar seus estudos
🦫 🤳🏻 Os influencers que ganham fama exibindo animais selvagens como pets
Tarifaço prejudica Estados 'bolsonaristas' 📈
Quem é Dona Maria, a personagem de IA que viraliza falando mal de Lula e do STF
Ibama X influenciadora: a briga na Justiça pela guarda da jaguatirica Pituca 🐯⚠️
Entrevista de emprego feita por IA? 💼📲
Os bilionários que aparecem nos novos arquivos do caso Epstein — e qual era a atuação dele no Brasil
Dá para identificar texto gerado por inteligência artificial?
Peixe gigante da Amazônia vira bolsa de luxo de $ 8 mil. Quanto ganham os pescadores? 🎣💰
Demitido do Itaú por produtividade baixa no home office: 'Trabalhei sete dias seguidos' 💻🔍
I was interviewed by Bellingcat to discuss how to use public records and data to investigate environmental crimes in Brazil. Held in English.
Panel discussion on investigating environmental crimes at GIJC 2023 in Gothenburg, Sweden.
Discussion about transparency and investigative journalism in Brazil.
Podcast discussion about corporate expenses investigation.
Presentation about DataFixers.org at the Lisbon International Fair.
Interview about air charter investigation and transparency in government spending.
Instituto Ayrton Senna
Research on transparency, state secrecy and prison expansion in Brazil, including work with Cambridge’s Prison Consensus Project.
Examines how claims about jobs, climate action and investment help justify and sustain Brazil’s prison expansion.
Read the article ↗Shows how weak oversight of classified information allows excessive secrecy and limits democratic accountability in Brazil.
Read the article ↗Explores how failures in Brazil’s prison system create openings for further expansion and mass incarceration.
Examines prison planning and the consequences of the policies that shape it.
Shows how access barriers and heavy redactions keep Brazilian military documents effectively secret even after declassification.
Read the thesis (in Portuguese) ↗
Instituto Ling