AI Knowledge Graph

A personal knowledge base covering AI and Cloud concepts, built with Obsidian.

Knowledge Graph

Open the interactive Knowledge Graph Explorer — download and open Graph/graph-explorer.html in a browser to browse notes by category, with collapsible hub nodes.

Contents

AI Concepts

• RAG — Retrieval-Augmented Generation
• Agentic RAG — RAG with autonomous agent workflows
• GraphRAG — Graph-based Retrieval-Augmented Generation
• RAG Evals — Evaluating RAG pipelines
• Hybrid Retrieval — Combining sparse and dense retrieval methods
• Embeddings — Vector embeddings and semantic representation
• Vector Database — Vector storage and similarity search
• Knowledge Bases — Knowledge base construction and management
• Long Context — Long context window techniques and trade-offs
• Fine-Tuning — Model fine-tuning techniques
• PEFT — Parameter-Efficient Fine-Tuning
• LoRA — Low-Rank Adaptation
• LoRA & QLoRA — Low-Rank Adaptation and Quantized LoRA
• Prompt Engineering — Prompting strategies
• Chain of Thought (CoT) — Step-by-step reasoning prompting strategy
• ReAct — Reasoning and Acting prompting framework
• Prompt Chaining — Sequential prompt workflows with gate checks
• RAG vs Fine-Tuning vs Prompt Engineering vs Long Context — Comparison guide
• Model Parameters — Key model parameters and their effects
• Top K — Top-K sampling for controlling output diversity
• Top P — Nucleus sampling for controlling output randomness
• Temperature — Temperature and its impact on model output
• Models — Overview of AI models and their capabilities
• Prompt Evaluations — Evaluating and scoring prompt quality
• AI Agent — Perception-reasoning-action loop and agent components
• Agentic Design Patterns — Core architectural patterns for agentic systems
• Agent Development Kit (ADK) — Google's code-first framework for building agents
• LangChain — Framework for building LLM-powered applications

Cloud Concepts

• AWS — Amazon Web Services overview
• Bedrock Services — AWS Bedrock AI services
• Amazon Bedrock Prompt Caching — Prompt caching on AWS Bedrock
• Bedrock Data Automation — Automating data pipelines with Bedrock
• Bedrock Fine-Tuning — Fine-tuning models on AWS Bedrock
• Bedrock Guardrails — Safety and guardrails on AWS Bedrock
• Converse API — Bedrock Converse API
• SageMaker — AWS SageMaker
• SageMaker Data Wrangler — Data preparation and feature engineering
• CloudWatch — Monitoring and observability on AWS
• Glue — Serverless data integration and ETL on AWS
• Amazon Transcribe — Automatic speech recognition service on AWS
• Amazon Polly — Text-to-speech service on AWS
• Amazon EC2 — Resizable virtual server compute on AWS
• Amazon Comprehend — Natural language processing service on AWS
• Amazon OpenSearch — Search and analytics service on AWS

System Design

• RAG Architecture — End-to-end RAG system architecture
• RAG Components — Components, embedding, retrieval & re-ranking, pipeline algorithms
• Model Parameters — Temperature ranges and parameter visualizations
• Prompt Evaluations — Prompt evaluation diagrams
• Chain of Thought (CoT) — CoT prompting diagram

Usage

Clone the repo and open in Obsidian to explore the interactive knowledge graph with all linked concepts.