Training models one step at a time
I'm a Research Scientist at Salesforce (joined via Tenyx acquisition). I currently work on the fundamentals of Large Language Models, focusing on training methodologies, understanding, and agentic applications.
What's New
- Apr 2026Paper out on arXiv: User Turn Generation as a Probe of Interaction Awareness in Language Models
- Feb 2026Paper out on arXiv: Interaction Theater: A case of LLM Agents Interacting at Scale
- Dec 2025Blog post with Silvio on the Agent-to-Agent Semantic Layer
- Nov 2025Paper out on arXiv: Echoing: Identity Failures when LLM Agents Talk to Each Other
- Nov 2025Paper on arXiv: Convergence dynamics of Agent-to-Agent Interactions with Misaligned objectives
Selected Publications
View allS Shekkizhar · arXiv Preprints, 2026
We propose user-turn generation as a probe of interaction awareness: given a conversation context of user query and assi...
S Shekkizhar, R Cosentino, A Earle, S Savarese · arXiv Preprints, 2025
As large language model (LLM) based agents interact autonomously with one another, a new class of failures emerges that ...
R Cosentino, S Shekkizhar · arXiv Preprints, 2024
The advancement of large language models (LLMs) for real-world applications hinges critically on enhancing their reasoni...