The Fifth Elephant 2025 Annual Conference CfP

The Fifth Elephant 2025 Annual Conference CfP

Speak at The Fifth Elephant 2025 Annual Conference

Suman Paul Choudhury

Suman Paul Choudhury

@sumanpc

Demystifying LLM Agents: Building Blocks, Workflows, and Best Practices

Submitted May 15, 2025

Talk description :

The rise of agentic workflows marks a significant shift in how we harness the capabilities of Large Language Models (LLMs) to perform complex, multi step reasoning and decision making tasks. Unlike traditional pipelines, where prompts are static and behavior is tightly scripted, agent based systems offer dynamic orchestration. LLMs can choose tools, recall memory, decompose problems, and iterate toward a goal.

This talk will explore the design patterns, trade-offs, and practical considerations involved in building such systems. We will introduce the concept of the augmented LLM, which extends base model capabilities with retrieval, tool use, and memory, forming the foundation for intelligent agents. We will also examine structured workflows such as prompt chaining, routing, and orchestrator-worker architectures, and analyze when each pattern is most effective.

Equally important are the good practises that favors simplicity over fragile complexity. Through hands on examples and architectural insights, this session will distill the principles behind agentic reliability, adaptability, and maintainability.

By the end of the session, attendees will have a grounded understanding of how to design and iterate on LLM agents that reason, act, and improve over time without sacrificing control or interpretability.

Takeaways:

  • Discover how agentic workflows enable LLMs to plan, use tools, recall memory, and act autonomously beyond simple prompts.
  • Explore practical design patterns and protocols from prompt chaining to multi-agent coordination—that power real-world LLM systems.
  • Learn practical strategies to build agent systems that deliver real world impact

Audience segment : Data Scientists , Applied Scientist , ML Engineering

Speaker Bio :

Suman Paul Choudhury is a Senior Data Scientist at Sahaj with over 8 years of industry experience in advanced data science and AI. He holds a Master’s degree in Signal Processing and specialises in computer vision, NLP, generative AI, and deep learning. Throughout his career, he has developed and deployed a wide range of AI solutions tailored to diverse client needs, driving measurable business impact.

Previous talks:

https://www.youtube.com/watch?v=HxRDyMB87dI&t=166s
https://www.youtube.com/watch?v=okOqiAJmaEo&list=PLarGM64rPKBnvFhv7Zgvj2t_q399POBh7&index=25&t=1s

https://www.linkedin.com/in/suman-paul-choudhury-22101319/
https://scholar.google.com/citations?user=7_CjdjwAAAAJ&hl=en&oi=ao

Comments

{{ gettext('Login to leave a comment') }}

{{ gettext('Post a comment…') }}
{{ gettext('New comment') }}
{{ formTitle }}

{{ errorMsg }}

{{ gettext('No comments posted yet') }}

Hosted by

Jump starting better data engineering and AI futures