Understanding LLMs
How large language models work and where they fit in modern applications.
AI • GENERATIVE AI • AGENTIC AI
Exploring how AI becomes a real system.
I’m exploring AI from the model level through production architecture — understanding not only how LLMs work, but how to design, evaluate, secure, scale, and operate AI systems in the real world.
01 — LEARNING SERIES
Learning AI in public.
A practical path from LLM fundamentals to production-ready enterprise AI and Agentic AI systems.
LLM vs RAG vs Fine-Tuning
When to use a model as-is, when to retrieve knowledge, and when to specialize behavior.
RAG Architecture
Embeddings, retrieval, hybrid search, and the path toward production knowledge systems.
Agentic AI
Agents, tools, planning, memory, orchestration, and intelligent workflows.
Production AI
Evaluation, observability, security, reliability, cost, and operating AI at scale.
AI Agents & Tool Calling
How agents interact with tools, APIs, enterprise systems, and controlled actions.
02 — AI ARCHITECTURE
How I think about AI systems.
AI architecture is more than selecting a model. The real system spans knowledge, orchestration, data, infrastructure, and the controls required to make it reliable.
Experience
AI applications • Assistants • Intelligent workflows
Models
LLMs • Generation • Reasoning • Specialized behavior
Knowledge
RAG • Embeddings • Vector search • Graphs • Enterprise data
Agents
Planning • Tools • Memory • Orchestration • Human approval
Platform
APIs • Events • Cloud • Data platforms • Distributed systems
Production
Evaluation • Security • Observability • Reliability • Cost
03 — AGENTIC AI
From answers to actions.
I’m exploring how systems move from generating a response to reasoning through a task, using tools, retrieving context, taking controlled actions, and coordinating workflows.
Reason
Understand goals, constraints, context, and the next best step.
Retrieve
Bring in trusted knowledge from documents, APIs, databases, and memory.
Act
Use tools and enterprise services to perform controlled actions.
Observe
Evaluate outcomes, capture telemetry, and keep the system accountable.
Building AI systems, not just AI demos.
The focus is practical: architecture choices, trade-offs, evaluation, security, reliability, and production readiness.
04 — BUILDING
From concepts to experiments.
The best way to understand AI architecture is to build it, measure it, and understand where the abstractions break.
AI / CLOUD
AI Cloud Analysis
Exploring AI-assisted cloud and Kubernetes analysis, including intelligent log analysis, orchestration, and production AI workflows.
View GitHub →RAG / ENTERPRISE AI
Enterprise Knowledge Assistant
A practical exploration of retrieval, context, model selection, evaluation, and the architecture required for enterprise knowledge assistants.
AGENTIC AI
Agentic Workflow Platform
Exploring agents, tool calling, orchestration, memory, events, and controlled actions across enterprise systems.
05 — WRITING & RESEARCH
Questions worth exploring.
I write about the architectural decisions behind AI systems, not just the technologies themselves.
Vector DB vs Graph DB
When should enterprise knowledge be represented as semantic similarity versus explicit relationships?
RAG vs Fine-Tuning
Should the system retrieve knowledge at runtime or change the model’s behavior?
Synchronous vs Event-Driven Agents
When should an agent wait for a response and when should it publish work into an event-driven flow?
Agent Memory
What belongs in short-term context, persistent memory, retrieval systems, or external state?