17 free courses, no signup wall
Architect-led enterprise cloud, security & AI
Fixed-price engagements, scoped on a discovery call
Skip to content

Knowledge hub

The Citadel knowledge graph

A connected reference for enterprise AI, cloud, and security engineering. Governed AI agents, RAG, LLMOps, Zero Trust, and the platform work underneath. Every node links to the next, so you can trace a problem from strategy to implementation.

Start with the AI blueprintBook an architecture review
In brief

What this knowledge hub covers, and who it is for

The Citadel knowledge graph is a connected reference hub for enterprise AI, cloud, and security engineering, written for the architects, engineers, and executives who have to get AI systems past a security review and into production. It is organised into five clusters. Enterprise AI covers governed AI agents, agentic AI, the Model Context Protocol, AI governance, retrieval against fine-tuning, LLMOps against MLOps, and agent frameworks. Security covers enterprise AI security, agent threats such as prompt injection and excessive agency, zero trust for agents and model endpoints, and cloud security. Cloud covers AWS, Azure, Google Cloud, Kubernetes, and Terraform. Development covers production RAG pipelines, vector databases, prompt engineering, MLOps, and data engineering for AI. Business covers engagement models, cloud cost, and return on investment. Every node links to the next, so a question can be traced from strategy through architecture to implementation without leaving the hub.

Start from your role

16 tracks, grouped by what you are accountable for rather than by technology. Each one carries a career ladder, the stack the work is done with, a reference project structure, and the primary sources behind it.

Build

Design and write the system.

  1. ArchitectAI ArchitectDesign AI systems that survive a security review, a budget cycle, and a year in production.12 guides · 4 repos
  2. EngineerAI EngineerBuild the retrieval, prompting, and evaluation machinery that turns a model into a product.12 guides · 4 repos
  3. LLM engineerLLM EngineeringPrompting, context, embeddings, and evaluation, the craft of making a model behave predictably.3 repos
  4. Agent engineerAI Agent EngineeringAgents are principals, not features. Each one needs an identity, a scope, and an audit trail.12 guides · 7 repos
  5. Data engineerAI Data EngineeringThe retrieval corpus is the product. Pipelines, vector stores, and the lineage behind them.5 repos

Run

Operate it once it is real.

  1. Platform engineerAI Platform EngineeringThe runtime every model lands on: GPUs, serving, autoscaling, and the bill at the end of it.5 repos
  2. Platform / SRELLMOps & MLOpsWhat changes operationally when the artifact is a model, and what does not.9 guides · 3 repos
  3. Infrastructure engineerAI InfrastructureKubernetes, Terraform, and GPUs, the substrate every AI system eventually lands on.12 guides · 7 repos
  4. DevSecOps engineerAI DevSecOpsSecure delivery for AI systems: pipelines, policy as code, secrets, and supply chain.5 repos

Govern

Prove it is safe to run.

  1. Security engineerAI SecurityThe model is rarely the vulnerability. The credentials, the retrieval corpus, and the tool surface are.12 guides · 5 repos
  2. Governance leadAI GovernanceGovernance is the artifact you hand a risk committee, not a policy nobody can evidence.9 guides · 3 repos

Lead & deliver

Decide what gets built and get it adopted.

  1. Product managerAI Product ManagementDeciding what an AI system should do, and what it must refuse to do.2 repos
  2. Delivery leadAI Project & Programme DeliveryGetting an AI programme from approved to adopted, across teams that do not report to you.1 repo
  3. Consultant / SEAI Consulting & Solution SalesDiagnosing what an organisation should build, and being trusted enough to say "not yet".3 repos
  4. Lead / Director / CAIOAI LeadershipBuilding the team, the operating model, and the mandate that lets AI work reach production.3 repos

Career

Get hired, get promoted, get paid.

  1. Any levelAI & Cloud CareerCertifications, salary data, and interview material, what actually moves a cloud or AI career.12 guides · 4 repos

Where should you start?

The graph has no required entry point, but three routes through it come up most often. Each one is three nodes deep and ends somewhere you can act on.

If you are an architect

Begin with the identity, credential, and autonomy model that has to clear a security review, then apply least privilege to agents and model endpoints, then settle the retrieval layer underneath.

  1. Governed AI Agents Identity, scoped credentials, tiered autonomy, and audit for agents in production.
  2. Zero Trust for AI Never-trust, least-privilege applied to agents and model endpoints.
  3. Vector Databases for RAG pgvector, Pinecone, Weaviate, and Milvus compared for the grounding layer.

If you are an engineer

Begin with how retrieval is actually built and operated, then compare the orchestration frameworks before committing to one, then settle the container and cluster question underneath the whole system.

  1. Production RAG Pipelines A senior engineer’s guide to running retrieval in production.
  2. Agent Frameworks LangChain, LlamaIndex, and AutoGen compared for enterprise use.
  3. Docker and Kubernetes When to use each, with real examples.

If you are an executive

Begin with what actually gets AI approved for production, then the reference path from pilot to governed system, then the engagement models and pricing behind delivery.

  1. AI Governance The engineering framework that gets AI approved for production.
  2. Enterprise AI Blueprint A reference path from AI pilot to governed production system.
  3. Enterprise Services Cloud, security, and AI transformation engagements, and what they cost.

Enterprise AI

AI architecture & agents

How governed AI agents, retrieval, and LLM operations actually ship in production, the decisions that survive a security review.

Governed AI Agents

Identity, scoped credentials, tiered autonomy, and audit for agents in production.

Open

What Is Agentic AI?

Planning, tool use, memory, and autonomy, and the governance it demands.

Open

Model Context Protocol (MCP)

The open standard connecting agents to tools and data safely.

Open

AI Governance

The engineering framework that gets AI approved for production.

Open

AI Agents (SIAS)

The Software-in-a-Service catalog of governed enterprise agents.

Open

RAG vs Fine-Tuning

When retrieval beats fine-tuning, and when it does not.

Open

LLMOps vs MLOps

What changes operationally when the model is an LLM.

Open

Agent Frameworks

LangChain, LlamaIndex, and AutoGen compared for enterprise use.

Open

Enterprise AI Blueprint

A reference path from AI pilot to governed production system.

Open

Security

AI & cloud security

Securing LLMs, data pipelines, model endpoints, and the cloud they run on, Zero Trust applied to AI systems.

Enterprise AI Security

Protecting LLMs, data pipelines, and model endpoints.

Open

AI Agent Security

Threats (prompt injection, excessive agency) and the controls that stop them.

Open

Zero Trust for AI

Never-trust, least-privilege applied to agents and model endpoints.

Open

Cloud Security

Identity, compliance, and zero-trust for cloud workloads.

Open

Security Practice

How Citadel approaches DevSecOps and AI security engagements.

Open

Zero Trust Implementation

A complete multi-cloud Zero Trust implementation guide.

Open

Kubernetes Security

Lessons from securing production clusters.

Open

AI Security Architecture

Architecture for protecting LLMs and their data.

Open

Cloud

Cloud & platform engineering

AWS, Azure, GCP, Kubernetes, and Terraform, the platform layer under every AI system.

AWS

Services, architecture patterns, and certification prep.

Open

Azure

Azure infrastructure, DevOps, and security.

Open

Google Cloud

GCP core services, networking, and data engineering.

Open

AWS vs Azure vs GCP

An architect’s honest comparison for 2026.

Open

Terraform / IaC

Infrastructure as Code from first principles.

Open

Docker & Kubernetes

When to use each, with real examples.

Open

Development

Building AI systems

Production RAG pipelines, orchestration frameworks, and the engineering practices behind reliable AI features.

Production RAG Pipelines

A senior engineer’s guide to RAG in production.

Open

Vector Databases for RAG

pgvector, Pinecone, Weaviate, and Milvus, how to choose.

Open

RAG on AWS Bedrock

Building a production RAG pipeline on Bedrock.

Open

Prompt Engineering

A practical guide for engineers.

Open

MLOps in Production

Monitoring, drift detection, and retraining.

Open

Data Engineering for AI

The pipeline architecture every ML team needs.

Open

AI Engineering

The AI/ML engineering discipline and skill path.

Open

Business

Strategy, ROI & engagements

AI strategy, cost, and the engagement models that turn a pilot into governed production, and into ARR.

Enterprise Services

Cloud, security, and AI transformation engagements.

Open

Consulting

How Citadel scopes and delivers engagements.

Open

Managed AI Operations

Ongoing operation of governed AI systems.

Open

AI ROI Calculator

Estimate the impact of an AI agent on a workflow.

Open

Cloud Cost / FinOps

Save 40% on your AWS, Azure, or GCP bill.

Open

Architecture Review

Book a discovery call for a governed-AI architecture review.

Open

Where the graph leads

The reference nodes above describe how governed AI systems are built. These four pages describe what Citadel does about it, and what that costs.

Agent Assurance Assessment

The fixed-scope engagement that turns the reading above into an inventory, an evidence pack, and one governed agent running. Three weeks, $15,000, price published.

Open

The SIAS agent catalog

The catalogued agents themselves, grouped by domain, with the three autonomy tiers that decide what each one is allowed to do without a human in the loop.

Open

Security and governance posture

The control-by-control answer to a security review: identity, credentials, policy checks, audit trail, subprocessors, and what happens to your data at exit.

Open

Enterprise engagements

The delivery areas, the fixed-price commercial model, and the monthly retainers that ongoing governed AI operations run under.

Open