Manikandan Jayaprakash

Senior Data Engineer

[Manikandan Jayaprakash @ManikandanJP]

6/6 — Store & next step

The generated tags can be stored along with the document chunk:

{
"page": page_number,
"text": chunk_text,
"tags": tags
}

The bigger picture:

PDF → Chunk → LLM → Tags → Metadata → Vector Store → RAG

[Manikandan Jayaprakash @ManikandanJP]

Spent some time learning LangChain and built a small GenAI tag builder for PDFs.

The flow:
📄 PDF → 🔪 Chunk text → 🤖 LLM generates 3–5 tags → 🗂️ Store as JSON/vector data

Simple project, but a great way to understand how document processing + LLMs + RAG building blocks

[Manikandan Jayaprakash @ManikandanJP]

5/6 — Generate tags

Now I use Claude to generate 3–5 meaningful tags for each chunk:

from langchain_anthropic import ChatAnthropic
llm = ChatAnthropic(model="claude-sonnet-4-6")
prompt = f"""
Generate 3-5 meaningful tags for this text.
Return only the tags.
Text:

[Manikandan Jayaprakash @ManikandanJP]

4/6 — Split into chunks

Large documents need to be broken into smaller pieces.

from langchain.text_splitter import RecursiveCharacterTextSplitter
splitter = RecursiveCharacterTextSplitter(
chunk_size=1000,
chunk_overlap=100
)
chunks = splitter.split_documents(

[Manikandan Jayaprakash @ManikandanJP]

3/6 — Load the PDF
I used PyPDFLoader to read the PDF:
from langchain_community.document_loaders import PyPDFLoader

docs = PyPDFLoader("file.pdf").load()
Now the PDF content is available as LangChain documents.

[Manikandan Jayaprakash @ManikandanJP]

2/6 — Install
First, install the required packages:
pip install langchain langchain-community pypdf langchain-anthropic
This gives us the basic building blocks for PDF loading, text processing and Claude integration.

[Manikandan Jayaprakash @ManikandanJP]

1/6

Spent some time learning LangChain and built a small GenAI tag builder for PDFs.

The goal was simple:

📄 PDF → 🔪 Split → 🤖 Generate tags → 🗂️ Store metadata

Here’s how I built it 👇

[Manikandan Jayaprakash @ManikandanJP]

🔧 How I built it:

1️⃣ Load the PDF using PyPDFLoader

2️⃣ Split the document into smaller chunks using RecursiveCharacterTextSplitter

3️⃣ Send each chunk to an LLM using ChatAnthropic

4️⃣ Generate 3–5 meaningful tags for each chunk

5️⃣ Store the tags as JSON/dictionary or use

[Manikandan Jayaprakash @ManikandanJP]

1/6

Spent some time learning LangChain and built a small GenAI tag builder for PDFs.

The goal was simple:

📄 PDF → 🔪 Split → 🤖 Generate tags → 🗂️ Store metadata

Here’s how I built it 👇

[Manikandan Jayaprakash @ManikandanJP]

@sairahul1 Guide

[Manikandan Jayaprakash @ManikandanJP]

@pathik455 Interested

[RJ Balaji @RJ_Balaji]

You never know, We could be just a click away from saving a life..!

https://t.co/eTwZPdyAB5

#KavyaNeedsADonar #RJBalaji #ShareToSaveALife

[குலோத்துங்கன் @kulothu]

இந்தியா மூன்று பக்கம் ஊழலாலும்,
ஒரு பக்கம் கடனாலும் சூழப்பட்ட நாடு..

[Manikandan Jayaprakash @ManikandanJP]

@kulothu sema

[Narendra Modi @narendramodi]

Met President Park Geun-hye. We covered substantial ground & are committed to enhanced economic cooperation that will benefit our nations.