Chunking Text for RAG 🔥| Day 17 of 30 Days of AI Coding Challenge
Welcome to Day 17 of the 30 Days of AI Coding Challenge 🚀
You extracted documents yesterday… but RAG doesn’t work on raw text. Today, we convert customer reviews into search-ready chunks.
🧠 We will learn today
Why chunking is critical for accurate AI responses
How to transform raw customer reviews into RAG-ready text chunks
Two practical chunking strategies:
One review = one chunk (best for short text)
Overlapping chunks for longer content
How to store structured chunks in Snowflake for downstream AI use
🔥Why Chunking:
LLMs and embedding models have strict token limits. Chunking ensures text fits within these limits, allowing the model to process information without losing critical details.
Smaller, targeted chunks (e.g., paragraphs or semantic sections) make it easier to find the most relevant information, boosting retrieval efficiency.
🛠 Prerequisites
Completed Days 1–17 of the challenge (If not, check the earlier videos and complete them first.)
• 30 Days of AI Coding Challenge 🚀 | Build A...
Timecodes
0:00 - Introduction
0:13 - What you will Learn today
0:34 - What is Chunking & Why we need it
1:56 - Why AI Coding Challenge
2:07 - Demo of the LLM Model Comparison Tool
6:54 - Code walkthrough of LLM Model Comparison Tool
7:09 - Loading Customer Reviews from Day 16
7:49 - Chunking Processing Strategy
8:45 - Create Chunks
10:55 - Check if Chunk Table exists
11:12 - Replace Mode for Snowflake Table
11:22 - Save Chunks into Table
12:12 - Query Saved data from Chunk Table
12:24 - View Full Chunk text
13:20 - write_pandas in SnowPark
13:39 - Implementation of Chunking for RAG
14:35 - Outcome of Day 17 of AI Coding Challenge
15:01 - What you will learn tomorrow
👉 Resources
Day 17 challenge guide:
[https://30daysofai.streamlit.app/?day=17]
write_pandas: Writes a pandas DataFrame to a table in Snowflake and returns a Snowpark DataFrame object referring to the table where the pandas DataFrame was written to.
http://docs.snowflake.com/en/develope...
Official challenge announcement:
[https://discuss.streamlit.io/t/the-30...]
Code reference (all days):
[https://github.com/sudeepkumar10x/30D...]
---
💼🧑💻 Join Our Data Engineering Community
Get exclusive learning resources, updates, and discussions.
👉 https://chat.whatsapp.com/FBv72iezg9M...
👉 Upload your own strategy and see its impact on the accuracy and context
💬 Comment “DONE” once you finish Day 17
📹 Tomorrow, we’ll start with Embedding 🔥
#30DaysOfAI #Streamlit #Snowflake #AICoding #DataEngineering #SudeepKumar10x #AIChallenge #ChatBot #LLM #LLMComparison #LLMModels #RAG #embedding #chunking #chunk