PeopleofSuperintelligence

Researcher

Erik Arakelyan

Erik Arakelyan is a machine learning researcher whose work studies how models reason over knowledge graphs, natural language and images. He studied music before earning computer science and machine learning degrees in Armenia and London, then completed a PhD at the University of Copenhagen. His paper on complex knowledge-graph queries won an ICLR 2021 Outstanding Paper Award, while a study of fragile natural-language-inference predictions received an Outstanding Paper award at EACL 2024. After research roles at ARM, Amazon Alexa AI and Cohere, he joined NVIDIA in 2025, coauthored a 2026 study of whether small language models can self-correct, and is a senior researcher there in 2026.

Now at
NVIDIA
Badges

Timeline

  1. 2014–2019

    From music to machine learning

    Earlier, he studies music. From 2014 to 2018, he studies computer science at the American University of Armenia on a full governmental scholarship; from 2018 to 2019, he earns an MSc in Machine Learning at University College London and conducts research in the Machine Reading Group under Pasquale Minervini.

  2. 2020–2023

    Applied machine learning at ARM

    From November 2020 to October 2023, he works at ARM as an applied machine learning engineer and tech lead, leading model tailoring and optimization teams.

  3. 2021

    A paper on complex queries

    In November 2020, he coauthors Complex Query Answering with Neural Link Predictors with Daniel Daza, Pasquale Minervini and Michael Cochez. The method uses pretrained neural link predictors to answer complex queries over incomplete knowledge graphs and reports relative Hits@3 gains of 8% to 40%; the paper wins the ICLR 2021 Outstanding Paper Award.

  4. 2021–2025

    A Copenhagen PhD on reasoning

    From September 2021 to 2025, he pursues a PhD in machine learning at the University of Copenhagen with the CopeNLU research group and the Pioneer Centre for Artificial Intelligence, under Isabelle Augenstein and Pasquale Minervini. His thesis, Reasoning Inconsistencies and How to Mitigate Them in Deep Learning, develops methods for models that reason over knowledge graphs, natural language and images; he defends it in April 2025.

  5. 2023–2025

    Visiting research at Alexa and Cohere

    From October 2023 to January 2024, he works with Amazon's Alexa AI team on grounded, faithful and verifiable LLM reasoning. From 2024 to 2025, he is a visiting researcher at Cohere AI, working on grounded, faithful and verifiable LLM reasoning.

  6. 2024

    An EACL outstanding paper

    Semantic Sensitivities and Inconsistent Predictions: Measuring the Fragility of NLI Models, coauthored with Zhaoqi Liu and Isabelle Augenstein, receives an Outstanding Paper award at EACL 2024.

  7. December 2025

    LLM reasoning at DevFest Armenia

    In December 2025, he leads a session on LLM reasoning at DevFest Armenia in Yerevan.

  8. Since 2025

    Researcher at NVIDIA

    In 2026, he coauthors More Yap, Less Meaning: Uncovering Self-Improvement Behavior in Small Language Models with Marina Igitkhanian; the study tests whether small language models can spot flaws in their reasoning and use self-generated hints to correct initially wrong answers. His interests include AI reasoning, explainable AI and knowledge graphs. He joins NVIDIA as a machine learning researcher in 2025 and is a senior researcher there in August 2026.

Career

  1. Since 2025NVIDIAMachine learning researcher
  2. 2024–2025CohereVisiting Researcher
  3. 2023–2024AmazonVisiting Researcher
  4. 2020–2023ArmApplied Machine Learning Engineer and tech lead

Education

  1. 2025University of CopenhagenPhD
  2. 2019University College LondonMSc
  3. 2018American University of Armenia

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Erik Arakelyan. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. Google has the highest combined h-index among AI organizationsIncludes NVIDIA. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes NVIDIA. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes NVIDIA. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Erik Arakelyan yet. Send us a free-licensed photo.

  1. Erik Arakelyan osoblanco.github.io. Checked 22 October 2026.
  2. Erik Arakelyan | alphaXiv alphaxiv.org. Checked 22 October 2026.
  3. [2011.03459] Complex Query Answering with Neural Link Predictors arxiv.org. Checked 22 October 2026.
  4. Outstanding paper award at ICLR 2021: Complex Query Answering with Neural Link Predictors discoverylab.ai. Checked 22 October 2026.
  5. PhD defence by Erik Arakelyan – Københavns Universitet di.ku.dk. Checked 22 October 2026.
  6. Best Paper Awards - EACL 2024 2024.eacl.org. Checked 22 October 2026.
  7. Innovation Meets Nature: A Look Back at DevFest Armenia 2025 intech.am. Checked 22 October 2026.
  8. March 2025 in Research neuralnoise.com. Checked 22 October 2026.
  9. AUA Alumna Presents Research at ACL 2026 GEM Workshop in San Diego newsroom.aua.am. Checked 22 October 2026.