Arvindh Arun

I'm an AI researcher working on evaluating bottlenecks for closed-loop systems with LLMs, with a focus on long-horizon capabilities: forecasting real-world events, executing long tasks reliably, and harnesses that utilize memory better.

Most recently, I was a Research Scientist Intern at Sakana AI, working on pretraining Looped Transformers. Previously, I've also done research at/with Google DeepMind and JPMorgan Chase's Investment Banking division.

Alongside, I'm pursuing a PhD co-advised by Steffen Staab and Antonio Vergari as part of the ELLIS and IMPRS-IS PhD programs.

I enjoy technical and philosophical discussions about AI, so reach out.

News

Oct 2026FutureSim will be presented as a featured poster at the Frontier Data Summit.
Oct 2026Attending COLM in San Francisco 🇺🇸
Sep 2026My internship work on pretraining Looped Language models will appear as a Spotlight at LP4M @ NeurIPS 2026.
Sep 2026FutureSim accepted to NeurIPS 2026 (Evaluations & Datasets Track).
Jul 2026FutureSim won the Best Paper Award at Forecasting as a New Frontier of Intelligence @ ICML 2026.
May 2026Started as a Research Scientist intern at Sakana AI in Tokyo 🇯🇵
Apr 2026Attending ICLR in Rio 🇧🇷
Jan 2026Measuring Long Horizon Execution in LLMs accepted to ICLR 2026.
Nov 2025Attending EMNLP in Suzhou 🇨🇳
Oct 2025Our work on isolating and benchmarking long-horizon execution in LLMs accepted to MTI-LLM @ NeurIPS 2025.
Aug 2025SEMMA, a foundation model for KGs, to appear in the main track of EMNLP 2025.
Jul 2025Attending ICML in Vancouver 🇨🇦
Jun 2025Attending Y Combinator's AI Startup School in San Francisco 🇺🇸
May 2025Corrective Unlearning in GNNs accepted to ICML 2025.
Mar 2025Now part of the MPI IMPRS-IS PhD program.
Feb 2025Our work on controversial content detection won the Best Paper Award at BeyondFacts @ WWW 2025.
Aug 2024Joined the ELLIS PhD Program, advised by Steffen Staab (ELLIS Unit Stuttgart 🇩🇪) and Antonio Vergari (University of Edinburgh 🇬🇧).
Jun 2024Attending NAACL in Mexico City 🇲🇽 (virtually).
May 2024Graduated with a BTech + MS in Computer Science from IIIT Hyderabad 🇮🇳
Apr 2024Our work on network effects accepted to WOAH @ NAACL 2024.
Feb 2024Attended Google Research Week 2024 in Bangalore.
Jan 2024Started working with the InfoNet team at Google DeepMind.
Oct 2023Attended ECAI 2023 in Kraków 🇵🇱 and visited ETH Zürich 🇨🇭
Jun 2023Our work on imposing fairness constraints in GNNs accepted as an oral at ECAI 2023.
May 2023Joined JPMorgan's Corporate & Investment Bank as an AI/ML intern.
May 2022Research intern with Prof. Hong Cheng at the Chinese University of Hong Kong 🇭🇰
Apr 2022Our work on fraud detection on the Google Play Store to appear at ACM Hypertext 2022.
Aug 2021Our submission to the GermEval 2021 Shared Task ranked 6th globally.
May 2021Joined the Precog lab for my master's thesis with Prof. Ponnurangam Kumaraguru.

Selected Publications

What Breaks Depth Extrapolation in Looped Language Models?
Arvindh Arun, Tianyu Zhao, Kai Arulkumaran

Spotlight, Linguistic Principles for Foundation Models @ NeurIPS 2026
FutureSim: Replaying World Events to Evaluate Adaptive Agents
Shashwat Goel*, Nikhil Chandak*, Arvindh Arun*, Ameya Prabhu, Steffen Staab, Moritz Hardt, Maksym Andriushchenko, Jonas Geiping

NeurIPS 2026
Best Paper Award, Forecasting as a New Frontier of Intelligence @ ICML 2026

PDF Code
The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMs
Akshit Sinha*, Arvindh Arun*, Shashwat Goel*, Steffen Staab, Jonas Geiping

ICLR 2026

PDF Code
SEMMA: A Semantic Aware Knowledge Graph Foundation Model
Arvindh Arun, Sumit Kumar, Mojtaba Nayyeri, Bo Xiong, Ponnurangam Kumaraguru, Antonio Vergari, Steffen Staab

EMNLP 2025

PDF Code
A Cognac Shot to Forget Bad Memories: Corrective Unlearning in GNNs
Varshita Kolipaka, Akshit Sinha, Debangan Mishra, Sumit Kumar, Arvindh Arun*, Shashwat Goel*, Ponnurangam Kumaraguru

ICML 2025

PDF Code

* Equal contribution. Full list on Google Scholar.