PParthainrpsarathy.hashnode.dev·2d ago · 7 min readMulti-Agent AI Is a Distributed Systems ProblemA single AI agent can be surprisingly simple: a model, a prompt, perhaps retrieval, tools, and a memory store. It can look convincing in a demo. The difficulty changes when a product adds more agents.00
PParthainrpsarathy.hashnode.dev·Aug 10 · 13 min readFrom Loop Engineering to Graph EngineeringThis article takes you through the practical evolution of an AI workflow — from a simple agent loop that starts to fail under real-world demands to a structured graph designed for reliability and cont00
PParthainrpsarathy.hashnode.dev·Jul 28 · 5 min readUnderstanding the LLM Inference Workload: From Tokens to AttentionBased on insights from Mark Moyou, Senior Solutions Architect at NVIDIA Why LLM Inference Is Different Send a prompt to a large language model and you're doing something fundamentally different from t11K
PParthainrpsarathy.hashnode.dev·Jul 20 · 9 min readAI Evals Explained: How to Know If Your AI Agent Actually WorksA practical guide to how production AI teams systematically test, trace, and monitor their AI systems. Building an AI agent is the fun part. Knowing whether the thing you just built is actually any go12AK
PParthainrpsarathy.hashnode.dev·Jul 16 · 13 min readAgent AI System Design, Explained: A Builder's Guide to Production-Grade Agentic SystemsMany people assume an AI agent is just a large language model wrapped in a chat interface. That assumption survives exactly as long as your demo runs on a laptop. The moment you move to a real applica01S
