What Actually Happens Inside an AI Agent?
A simple mental model for understanding the loop behind modern AI agents.
AI Engineer · Tech Lead
I'm Snehal Agrawal — a Tech Lead and AI Engineer working at the intersection of AI systems, distributed architecture, and product engineering.
This is a collection of ideas, patterns, experiments, and lessons from building AI systems in the real world.
Understanding the building blocks behind modern AI systems — from LLMs and embeddings to inference and evaluation.
Designing useful LLM-powered applications that solve real problems instead of stopping at the demo.
Exploring agents, tools, workflows, memory, orchestration, and the engineering patterns behind agentic systems.
Reliability, evaluation, observability, performance, cost, security, and everything that matters after the prototype.
The engineering decisions, trade-offs, and system designs behind scalable production AI.
A simple mental model for understanding the loop behind modern AI agents.
Why good AI architecture often means knowing when not to use an LLM.
A practical way to think about deterministic logic, LLMs, and where each belongs in an AI system.
The goal
AI is moving fast. Good engineering still matters.