Software Engineering Radio - the podcast for professional software developers SE Radio 673: Abhinav Kimothi on Retrieval-Augmented Generation
Jun 18, 2025
Abhinav Kimothi, co-founder and VP at Yarnit and author of A Simple Guide to Retrieval-Augmented Generation, dives into the world of RAG. He elucidates key concepts like large language models and context windows while addressing issues such as hallucinations in AI. The discussion reveals essential components for building RAG systems, including prompt engineering and data chunking. Abhinav also shares insights on the trade-offs between open-source and proprietary models, plus the future potential of RAG, from multi-hop reasoning to agentic AI.
AI Snips
Chapters
Books
Transcript
Episode notes
Understanding Large Language Models
- Large language models (LLMs) are deep learning models trained on massive text using Transformers architecture.
- They predict the next word in a sequence based on probabilities, enabling natural language understanding and generation.
Prompt Engineering Importance
- Prompt engineering involves crafting effective natural language instructions to LLMs.
- The structure and logic in prompts significantly influence the quality of LLM outputs.
LLM Context Window Explained
- LLMs have a limited context window defining how much text they can process at once.
- Context window size has grown from small to hundreds of thousands of words in advanced models.
