PParthainrpsarathy.hashnode.dev·3d ago · 11 min readSecuring AI Agents in Production: A Practical Guide for First- and Second-Time BuildersThe One Rule Never put a security decision inside the prompt. A prompt is text, and a crafted input can override text. Permissions, redaction rules, and approval gates written into a prompt are sugges02AF
PParthainrpsarathy.hashnode.dev·Sep 1 · 14 min readChunking Is Design, Not Preprocessing: The Hidden Decision Behind Every RAG SystemRetrieval-Augmented Generation was meant to solve one simple problem: language models forget. RAG gave them a way to look things up — to ground their words in real knowledge. But grounding is fragile,11K
PParthainrpsarathy.hashnode.dev·Aug 27 · 13 min readHow KV Cache and Paged Attention Unlock Faster, Cheaper LLM InferenceTL;DR: LLM inference bottlenecks are usually memory, not compute. KV cache eliminates redundant attention computation. Paged attention eliminates memory fragmentation — cutting waste from ~60–80% down00
PParthainrpsarathy.hashnode.dev·Aug 20 · 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
