What the project did
Deepfake detection is a race between generators and detectors, and the detectors are expensive: the strongest systems are slow, which limits how much video anyone can screen. Tarr attacked the cost rather than the accuracy. “My project uses artificial intelligence to detect deepfake videos,” he said on winning, and it “speeds the process up by a factor of 10 while retaining state-of-the-art accuracy.”
His method was to study the top-performing entries from a million-dollar deepfake detection contest, refactor their code and train a custom face detector in place of the general-purpose components those systems relied on. The result was not a new claim about what deepfakes look like but a faster path to the same verdict — which, for anyone trying to screen video at scale, is the difference between a laboratory technique and a usable tool.
The competition he won
The BT Young Scientist and Technology Exhibition is Ireland's national science fair and one of the longest-running in Europe. The 2021 edition opened on 6 January and was held online because of the pandemic, with more than 500 student projects on display. Its portal recorded 3.5 million visits from 77 countries.
Tarr was named overall winner on 8 January. The prize was €7,500 and the right to represent Ireland at the EU Contest for Young Scientists. Leonard Hobbs of Trinity College Dublin, who chaired the technology judging panel, said Tarr had “demonstrated an expertise in computer science which was well beyond his years” and singled out “the level of coding he deployed in developing the extremely complex program.”
Five attempts
The detail that separates this from a lucky year is the count: 2021 was the fifth time Tarr had competed at the exhibition. A national science fair is one of the few arenas where a teenager can enter the same competition annually and be judged each time against a fresh field of several hundred projects.
By the time he won it, he was a self-taught programmer working in one of the most contested areas of applied machine learning, in a school in west Cork, during a year in which the exhibition itself could not be held in a building. Asked about what came next, he said he saw himself “almost certainly” going into entrepreneurship.
What is documented — and what is not
Firm: the win and its date, the school and county, the project's claim of a tenfold speed-up at state-of-the-art accuracy, the 150,000-plus lines of code, the €7,500 prize, the EU contest place, and the judge's assessment.
Open: no date of birth is stated in the source used here, so his age at the time is not given as a number. Nor does this record document what became of the software, whether it was published, licensed or built into a company, or what he studied afterwards. His stated intention to go into business is a statement of intent made in January 2021, not a record of what followed.
Why the result carried outside Ireland
Deepfake detection moved from a research curiosity to a policy problem over the years surrounding this project, as synthetic video became cheap enough to make and convincing enough to circulate. Systems that work only in a laboratory do not help; systems fast enough to be run over large volumes of uploaded video might.
That is the reason a school science project attracted attention beyond the exhibition hall. The claim was narrow, testable and aimed at the exact bottleneck the field had — and it came from a secondary-school student who had entered the same competition four times before and had written the whole thing himself.
| Person | Country | Milestone | Age / Stat |
|---|
What is documented — and what is not
✓ Documented
- Won the BT Young Scientist and Technology Exhibition on 8 January 2021 as a Bandon Grammar School student
- Project detects deepfakes ten times faster at state-of-the-art accuracy, in more than 150,000 lines of code
- Prize of €7,500 and a place at the EU Contest for Young Scientists
- Fifth time competing at the exhibition
? Unverified, disputed or wrong
- No date of birth or exact age in this source
- No record here of what happened to the software after 2021
- No documented university, company or later research in these sources
Questions people ask about Greg Tarr
What did Greg Tarr invent?
Software that detects deepfake videos roughly ten times faster than the leading systems while keeping their accuracy, built from more than 150,000 lines of code.
How did he do it?
He analysed the best entries in a million-dollar deepfake detection contest, refactored their code and trained a custom face detector to replace slower general-purpose components.
What did he win?
The BT Young Scientist and Technology Exhibition 2021, worth €7,500, plus a place representing Ireland at the EU Contest for Young Scientists.
How old is Greg Tarr?
The source used here describes him as a Leaving Certificate student in January 2021 but gives no date of birth, so no exact age is stated.
Where did he go to school?
Bandon Grammar School in County Cork.
Where is Greg Tarr now?
No source in this record documents what he has done since 2021. He said at the time that he saw himself “almost certainly” going into entrepreneurship.
Is Greg Tarr on Wikipedia?
No article was found for him. The strongest available source is the contemporaneous Silicon Republic report on his win.
How competitive is the BT Young Scientist exhibition?
The 2021 edition, held online, featured more than 500 student projects and drew 3.5 million visits to its portal from 77 countries.
Quick quiz: how well do you know Tarr’s story?
What did Greg Tarr's winning project do?
Which competition did he win in 2021?
How many lines of code did the project run to?
Sources
- Young Scientist winner finds improved method of detecting deepfakes — Silicon Republic, 2021-01-08
Checked by the Geniuses.club editorial team. Ages are given as of September 2026.
Discover More Child Prodigies
50+ stories of contemporary young geniuses changing chess, music, science, and art.
Explore All Prodigies →
