The Fifth Elephant 2025 Annual Conference CfP
Speak at The Fifth Elephant 2025 Annual Conference
Shruti Dhavalikar
@shrutidhavalikar
Submitted May 14, 2025
Agentic applications powered by large language models (LLMs) are transforming the way we build intelligent, interactive systems. Agentic applications powered by large language models (LLMs) are transforming the way we build intelligent, interactive systems. However, their non-deterministic nature introduces significant challenges when it comes to testing and evaluation. How can we ensure reliability, correctness, and consistent behaviour in systems that are inherently probabilistic?
This talk will dive into practical strategies for evaluating agentic applications, focusing on a conversational agent use case. Using a real-world use case, we dissect the application into key components—query understanding, data source orchestration, tool invocation, and response synthesis—and examine how each can be evaluated deterministically, even when built on stochastic foundations.
Key Takeaways
Target Audience
Speaker Bio
Shruti Dhavalikar is a seasoned Data Scientist with over 6 years of experience, currently serving as a Solution Consultant at Sahaj Software in Pune, India. Her expertise lies in transforming complex data into actionable insights, delivering end-to-end product cycles under Agile methodologies, and ensuring scalable, clean, and robust coding across diverse tech stacks. Shruti’s strong communication skills have enabled her to effectively interact with clients, contributing to successful project outcomes. Beyond her professional pursuits, she harbours a curiosity for cosmology and space, and enjoys exploring different cuisines as a travelling foodie.
{{ gettext('Login to leave a comment') }}
{{ gettext('Post a comment…') }}{{ errorMsg }}
{{ gettext('No comments posted yet') }}