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 2026Attending COLM in San Francisco 🇺🇸
• 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

1. 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

   PDFCode

2. The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMs

   Akshit Sinha*, Arvindh Arun*, Shashwat Goel*, Steffen Staab, Jonas Geiping

   ICLR 2026

   PDFCode

3. SEMMA: A Semantic Aware Knowledge Graph Foundation Model

   Arvindh Arun, Sumit Kumar, Mojtaba Nayyeri, Bo Xiong, Ponnurangam Kumaraguru, Antonio Vergari, Steffen Staab

   EMNLP 2025

   PDFCode

4. 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

   PDFCode

* Equal contribution. Full list on Google Scholar.