diff --git a/churn/ReadMe.md b/churn/ReadMe.md
index 6539738..ea8fa8f 100644
--- a/churn/ReadMe.md
+++ b/churn/ReadMe.md
@@ -9,6 +9,14 @@ This project delivers an end‑to‑end Telecom Customer Churn Prediction soluti
 
 Both models are trained on rich, synthetic telecom datasets—incorporating usage patterns, billing details, and support interactions—to predict which customers are at highest risk of churn. By identifying high‑risk segments in advance, operators can deploy targeted retention offers and optimize churn‑prevention campaigns, ultimately safeguarding revenue and improving customer lifetime value.
 
+## Software Requirement
+
+| Software      | Version |
+|---------------|---------|
+| Python        | 3.12    |
+
+> Tested with Python 3.12 as the minimum version. Python 3.13 and later 3.x releases may work but have not been validated.
+
 ## Model Options
 
 Both the Balanced Random Forest and LightGBM classifiers are well‑suited for binary churn prediction, but they achieve this in different ways that are worth understanding:
diff --git a/revenueassurance/ReadMe.md b/revenueassurance/ReadMe.md
index 9b2a28f..9c6bbe5 100644
--- a/revenueassurance/ReadMe.md
+++ b/revenueassurance/ReadMe.md
@@ -2,6 +2,7 @@
 
 - [Revenue Assurance and Fraud Management (RAFM) with AI Assistance](#Revenue-Assurance-and-Fraud-Management-\(RAFM\)-with-AI-Assistance)
   - [Project Overview](#Project-Overview)
+  - [Software requirement](#Software-Requirement)
   - [Options](#Options)
   - [Data](#Data)
   - [Results](#Results)
@@ -24,6 +25,14 @@ The models are trained on synthetic telecom data to predict fraud cases and iden
 
 Data-Set: https://huggingface.co/datasets/fenar/revenue_assurance
 
+## Software Requirement
+
+| Software      | Version |
+|---------------|---------|
+| Python        | 3.12    |
+
+> Tested with Python 3.12 as the minimum version. Python 3.13 and later 3.x releases may work but have not been validated.
+
 ## Options 
 
 (1) Revenue Assurance is a domain where traditional machine learning models, such as tree-based methods, often perform very well. Specifically, models like the `Balanced Random Forest` (01-xxx Worx) can be advantageous due to their inherent ability to handle class imbalance and interpretability.<br>[ Test Accuracy: 0.99962] <br>
diff --git a/serviceassurance/ReadMe.md b/serviceassurance/ReadMe.md
index b3bbe45..876f0a7 100644
--- a/serviceassurance/ReadMe.md
+++ b/serviceassurance/ReadMe.md
@@ -7,6 +7,14 @@ This project aims to deliver a service assurance insights (net promoter score -N
 
 **🎥 Demo Video**: [Watch on YouTube](https://youtu.be/gFbtux0dGVA) <br>
 
+## Software Requirement
+
+| Software      | Version |
+|---------------|---------|
+| Python        | 3.12    |
+
+> Tested with Python 3.12 as the minimum version. Python 3.13 and later 3.x releases may work but have not been validated.
+
 ## Data
 ![Service Assurance Data Structure](https://raw.githubusercontent.com/fenar/etc-ai-wrx/main/serviceassurance/data/svcass-datainsp.png)<br>
 
diff --git a/starlink/README.md b/starlink/README.md
index c58ef8c..84419b8 100644
--- a/starlink/README.md
+++ b/starlink/README.md
@@ -12,6 +12,14 @@ To empower travelers with insights into Starlink's internet service quality at t
     <img src="https://raw.githubusercontent.com/tme-osx/TME-AIX/refs/heads/RedHat-Special/starlink/images/starling-qoe-moods.png" width="640"/>
 </div>
 
+## Software Requirement
+
+| Software      | Version |
+|---------------|---------|
+| Python        | 3.12    |
+
+> Tested with Python 3.12 as the minimum version. Python 3.13 and later 3.x releases may work but have not been validated.
+
 ## Data
 ![Data Structure](https://raw.githubusercontent.com/tme-osx/TME-AIX/refs/heads/main/starlink/images/starlink-data.png)<br>
 DataSet Location: https://huggingface.co/datasets/fenar/starlink