DevConf.IN 2025

Mathavan S G

Mathavan is an AI Engineer passionate about leveraging AI and data-driven insights to solve complex business challenges. With expertise in fine-tuning Large Language Models and creating high-quality datasets, he brings hands-on experience from his role as a Delivery Data Analyst at Turing. His work spans Generative AI, machine learning, and data visualisation, backed by a strong foundation in Python, Power BI, and cloud platforms like Azure. Explore his projects on GitHub: @MathavanSG.


Company or affiliation

Turing

Job title

Data Analyst


Session

03-01
13:00
35min
Enhancing Reasoning in SLMs: A Chain-of-Thought Fine-Tuning
Mathavan S G

This talk explores chain-of-thought reasoning, a fine-tuning technique to enhance the logical reasoning capabilities of SLMs like LLaMA (1B/3B parameters). Attendees will learn the full process—from dataset preparation to fine-tuning and evaluation—demonstrating how smaller models can deliver interpretable, step-by-step responses with minimal resources.

Key Takeaways
  • Fine-tune small language model for better reasoning and interpretability.
  • Practical insights on datasets, training, and hardware.
  • Apply scalable techniques to open-source SLMs.
Target Audience

AI&ML engineers, data scientists, and researchers seeking to enhance reasoning in small-scale models with practical, resource-efficient methods.

Why Attend

Learn actionable techniques to make SLMs smarter, more interpretable, and accessible for real-world applications.

AI, Data Science, and Emerging Tech
Raigad Room (Chanakya Building / School of Business)