Debjyoti Paul - Learning to Act Reinforcement Learning for Agentic LLM Systems Debjyoti Paul - Learning to Act Reinforcement Learning for Agentic LLM Systems

Large Language Models (LLMs) have demonstrated impressive reasoning and generation abilities, but building agentic systems—AI that can plan, use tools, interact with environments, and achieve goals autonomously—requires more than prompting. A key challenge is enabling these systems to learn how to act, not just how to respond. This talk explores how Reinforcement Learning (RL) can transform LLMs into effective decision-making agents. We examine the architecture of modern agentic systems where LLMs serve as planners and reasoning engines, while RL provides the feedback loop that enables continuous improvement through interaction with tools, APIs, and external environments.

The session will walk through practical design patterns for integrating RL with LLM-based agents, including task decomposition, action selection, tool execution, and reward shaping. We will discuss how RL techniques such as policy optimization and reward modeling can help agents improve planning, reduce hallucinations, and learn reliable strategies for complex multi-step tasks.

Using concrete examples—from automated workflows to multi-step information retrieval and decision-making systems—we illustrate how RL-driven feedback can improve agent performance over time. We also discuss common challenges, including reward design, exploration, stability, and evaluation of agent behavior.

By the end of the talk, attendees will gain a practical understanding of how to design self-improving agentic AI systems that combine the reasoning capabilities of LLMs with the learning dynamics of reinforcement learning.

Debjyoti is a Data Scientist at Amazon with over 9 years of industrial experience in Natural Language Processing (NLP), Large Language Models (LLMs), and Agentic AI and Responsible AI, Currently Debjyoti is Leading Agentic development and actively working on Agent learning primarily focusing on Agentic System improvement from Context Engineering to RL based Learning framework. Prior to this Debjyoti has led AI-driven solutions in enterprise applications focusing on Anomaly Detection, Recommendation system, NLP, Information extraction and Computer Vision. Their research expertise spans AI ethical AI governance, bias mitigation, and scalable LLM deployment, model interpretability ensuring responsible AI adoption across industries. With hands-on experience in developing production-grade AI systems, they actively research fairness, robustness, and transparency in AI, contributing to frameworks that enhance trust and accountability in AI-driven decision-making.

This session is brought to you by the Cohere Labs Open Science Community - a space where ML researchers, engineers, linguists, social scientists, and lifelong learners connect and collaborate with each other. We'd like to extend a special thank you to Katrina Lawrence and Neel Ghoshal, Leads of our ML Math group for their dedication in organizing this event.

If you’re interested in sharing your work, we welcome you to join us! Simply fill out the form at https://forms.gle/ALND9i6KouEEpCnz6 to express your interest in becoming a speaker.

Join the Cohere Labs Open Science Community to see a full list of upcoming events (https://tinyurl.com/CohereLabsCommunityApp).