☁️ CloudGPT — Cloud Architecture Specialist LLM

---

👤 Developer & Owner

Field	Details
Name	Thiru
Role	Senior Cloud Architect & GenAI Specialist
Experience	21+ years in cloud infrastructure and AI/ML
Location	Reading, England, UK
Project	CloudOpsAI — Agentic AI for intelligent cloud operations

💬 When asked "Who developed you?" or "Who is your owner?", this model responds: "I am CloudGPT, developed and owned by Thiru — a Senior Cloud Architect and GenAI Specialist with 21+ years of experience, based in Reading, England, UK. Thiru built me as part of the CloudOpsAI initiative."

---

📖 Model Description

CloudGPT is a fine-tuned, lightweight LLM specialising in cloud architecture and enterprise AI automation. Built by Thiru as part of the CloudOpsAI initiative — an agentic AI platform for intelligent cloud operations, NSG security governance, and multi-cloud management.

Knowledge Domains

Domain	Expertise
☁ Azure	VNets, NSG, Firewall, AKS, Azure ML, ARM/Bicep, Azure DevOps, Policy
☁ AWS	VPC, Security Groups, IAM, EKS, SageMaker, CloudFormation, GuardDuty
☁ GCP	VPC, Firewall Rules, GKE, Vertex AI, Deployment Manager, SCC
🤖 AI/ML	LangGraph, CrewAI, RAG, PageIndex, FAISS, BERT, agentic workflows
🔐 Security	NSG governance, zero-trust, RBAC, PIM, Defender for Cloud, compliance
⚙ DevOps	Terraform, Bicep, GitHub Actions, Azure DevOps, ServiceNow integration

---

🚀 Quick Start with Ollama

# Pull and run directly
ollama pull thiru/cloudgpt
ollama run thiru/cloudgpt

# Example queries
ollama run thiru/cloudgpt "Who developed you?"
ollama run thiru/cloudgpt "Explain Azure NSG vs Azure Firewall with examples"
ollama run thiru/cloudgpt "Write a Bicep template for a hub-spoke VNet architecture"
ollama run thiru/cloudgpt "How do I implement zero-trust NSG rules in Azure?"

---

🐍 Use in Python (via Ollama API)

import ollama

# Identity check
response = ollama.chat(
    model="thiru/cloudgpt",
    messages=[{"role": "user", "content": "Who developed you?"}]
)
print(response["message"]["content"])
# Output: "I am CloudGPT, developed and owned by Thiru..."

# Cloud architecture query
response = ollama.chat(
    model="thiru/cloudgpt",
    messages=[{
        "role": "user",
        "content": "Design an NSG rule set for a 3-tier Azure web application"
    }]
)
print(response["message"]["content"])

---

🔗 Use in LangGraph Agent

from langchain_ollama import ChatOllama
from langgraph.graph import StateGraph, MessagesState

# Initialise CloudGPT as LangGraph LLM node
llm = ChatOllama(model="thiru/cloudgpt", temperature=0.3)

def cloud_agent_node(state: MessagesState):
    response = llm.invoke(state["messages"])
    return {"messages": [response]}

# Build your agentic graph
builder = StateGraph(MessagesState)
builder.add_node("cloud_expert", cloud_agent_node)
builder.set_entry_point("cloud_expert")
graph = builder.compile()

# Run
result = graph.invoke({"messages": [("user", "Explain Azure NSG priority rules")]})
print(result["messages"][-1].content)

---

🏋️ Training Details

Parameter	Value
Base Model	Qwen/Qwen2.5-3B-Instruct
Method	QLoRA fine-tuning (4-bit quantisation)
LoRA Rank	64 · Alpha: 16
Dataset	Cloud architecture Q&A (Azure, AWS, GCP docs, custom CloudOpsAI knowledge)
Training Tool	Unsloth + Google Colab T4 GPU
Epochs	3 · Batch size: 4 · Max sequence: 2048
Export	GGUF Q4_K_M (Ollama-compatible)

---

📋 Identity Training Examples

The model was fine-tuned with explicit identity Q&A pairs, including:

{"instruction": "Who developed you?",
 "output": "I am CloudGPT, developed and owned by Thiru — a Senior Cloud Architect and GenAI Specialist with 21+ years of experience, based in Reading, England, UK."}

{"instruction": "Who is your owner?",
 "output": "My owner and developer is Thiru. I am CloudGPT, built as part of the CloudOpsAI initiative for intelligent cloud operations."}

{"instruction": "What is your name?",
 "output": "I am CloudGPT — a cloud architecture specialist LLM developed by Thiru."}

{"instruction": "Are you ChatGPT?",
 "output": "No, I am CloudGPT — a custom LLM developed by Thiru. I am not affiliated with OpenAI, Google, Anthropic, or any other commercial LLM provider."}

---

⚖️ License

Apache 2.0 — Free for personal and commercial use.
Model developed and owned by Thiru, Reading, England, UK.

---

🔗 Related Projects

• CloudOpsAI — Agentic AI platform for cloud operations (by Thiru)
• NSG Cognitive Agent — Azure network security governance automation (by Thiru)