Understanding AI Agents From an Engineer's Perspective
A practical introduction to how AI agents work, including tools, memory, reasoning, workflows, and orchestration.
Hello, I'm Haseeb Sultan.
AI Platform
Engineer
I build and learn about AI platforms, agentic systems, LLMOps, cloud infrastructure, Kubernetes, MCP, and AI security.
I'm documenting what I learn, what I build, and the engineering lessons along the way.
Currently Exploring
- Agentic AI
- LLMOps
- MCP
- AI Infrastructure
- Kubernetes for AI
- AI Security
- RAG
- AI Agents
Notes from the workbench
Featured Writing
Running AI Workloads on Kubernetes
Exploring the infrastructure challenges and architecture patterns behind running production AI systems.
Building Secure MCP-Based AI Agents
Notes on designing secure Model Context Protocol architectures for enterprise AI agents.
Putting ideas into practice
Selected Projects
Rocotal
Enterprise AI security and governance platform focused on protecting interactions between users, AI agents, and large language models.
Secure Document Agent
An AI agent architecture for securely searching, reading, and interacting with enterprise documents using MCP.
AI Infrastructure Explorer
An experiment around securely understanding and visualizing cloud infrastructure using AI-assisted workflows.
A foundation to build on
From Infrastructure
to AI Platforms
My background started with DevOps, cloud infrastructure, Kubernetes, and platform engineering. As AI systems move into production, I've become increasingly focused on the infrastructure, security, orchestration, and operational challenges behind them. Today, I'm exploring the intersection of AI engineering and platform engineering.
Where I focus
What I Work On
AI Platforms
Designing infrastructure and platforms for running AI applications and agentic systems reliably.
Agentic Systems
Exploring AI agents, tool use, workflows, orchestration, MCP, and multi-agent architectures.
LLMOps
Deployment, observability, evaluation, reliability, scalability, and lifecycle management of LLM applications.
AI Security
Prompt security, guardrails, MCP security, data protection, and secure enterprise AI architectures.
A growing collection, not a finished story
Learning in Public
AI is evolving quickly, and I'm learning by building. I use this website to document concepts, experiments, architecture decisions, mistakes, and lessons that I want to remember — and that may help other engineers along the way.
Explore My Writing