I help teams turn AI agent prototypes into systems they can bill customers for — evals, cost control, and observability. 9 years of distributed systems, now applied to agent reliability.https://durgeshrathod.com/en-us© 2026 Durgesh Rathodhttps://durgeshrathod.com/notes/agent-design-review/https://durgeshrathod.com/notes/agent-design-review/Reviewing an agent feature before it ships is the cheapest reliability work available, and most teams review the prompt rather than the architecture. Nine questions that catch the expensive problems while they are still free.Sun, 02 Aug 2026 00:00:00 GMTNotesagent design reviewAI code review checklistLLM architecture reviewagent feature reviewhttps://durgeshrathod.com/notes/context-window-budget/https://durgeshrathod.com/notes/context-window-budget/A million-token window does not mean you should use it. Decompose one real conversation and the proportions are usually a surprise — history dominates, tool definitions are larger than expected, and the useful content is a minority of what you pay for.Sun, 02 Aug 2026 00:00:00 GMTNotesLLM context windowcontext window budgetconversation history costLLM token usagehttps://durgeshrathod.com/notes/event-driven-agents/https://durgeshrathod.com/notes/event-driven-agents/A 20% input spike once turned a healthy pipeline into an outage because it had no backpressure — it had optimism. Every concept that prevents that has a direct agent equivalent, and almost no agent codebase implements any of them.Sun, 02 Aug 2026 00:00:00 GMTNotesevent-driven agentsKafka LLM architectureasync agent designagent backpressureLLM streaming datahttps://durgeshrathod.com/notes/llm-as-judge-overused/https://durgeshrathod.com/notes/llm-as-judge-overused/It is slow, it costs money on every CI run, and it is non-deterministic — which means your test suite has flaky tests by construction. About 70% of the checks teams reach for a judge to make can be written as assertions instead.Sun, 02 Aug 2026 00:00:00 GMTNotesLLM as judgeLLM evaluation methodsAI eval assertionsLLM testing CIhttps://durgeshrathod.com/notes/mcp-in-production/https://durgeshrathod.com/notes/mcp-in-production/Model Context Protocol moves your tool boundary into a place where validation and authorisation naturally live. That is the real benefit, and it is not the one usually advertised. What it does not give you: security, cheaper tokens, or a reason to trust tool arguments.Sun, 02 Aug 2026 00:00:00 GMTNotesModel Context Protocol productionMCP best practicesMCP securityMCP tool designMCP vs function callinghttps://durgeshrathod.com/notes/model-routing/https://durgeshrathod.com/notes/model-routing/Routing tasks by difficulty typically recovers 40–70% of spend. It is also the only cost lever that can silently degrade quality — which is why the teams that most need it are usually the ones least equipped to do it safely.Sun, 02 Aug 2026 00:00:00 GMTNotesLLM model routingreduce LLM costsmodel selection by taskLLM cost optimisationhttps://durgeshrathod.com/notes/prompt-caching-vs-semantic-caching/https://durgeshrathod.com/notes/prompt-caching-vs-semantic-caching/They get discussed interchangeably and they solve different problems with different risks. One is a billing mechanism with a write premium that can lose you money; the other returns a previous answer to a different question and can leak across tenants.Sun, 02 Aug 2026 00:00:00 GMTNotesprompt cachingsemantic caching LLMLLM cache hit rateLLM cost cachinghttps://durgeshrathod.com/notes/prompt-versioning/https://durgeshrathod.com/notes/prompt-versioning/If a customer reports a bad answer from last Tuesday, can you reconstruct the exact prompt that produced it? Most teams cannot, and it makes every subsequent investigation archaeology.Sun, 02 Aug 2026 00:00:00 GMTNotesprompt versioningLLM prompt managementprompt rollbackLLM reproducibilityhttps://durgeshrathod.com/notes/tool-definition-token-cost/https://durgeshrathod.com/notes/tool-definition-token-cost/Every tool schema you attach is billed on every single call, whether the agent uses it or not — plus a per-model tool-use system prompt of 286 to 804 tokens. Here is how to measure that fixed cost and what it does to your bill at scale.Sun, 02 Aug 2026 00:00:00 GMTNotesLLM cost optimisationtool definitions tokensagent costprompt cachingMCPhttps://durgeshrathod.com/notes/why-ai-agents-fail-in-production/https://durgeshrathod.com/notes/why-ai-agents-fail-in-production/Almost every failing agent I have looked at is failing on engineering discipline, not model capability. The evidence is that upgrading the model rarely fixes it — and the four things that do are unglamorous.Sun, 02 Aug 2026 00:00:00 GMTNoteswhy AI agents fail in productionAI agent reliabilityLLM engineering disciplineproduction AI systemshttps://durgeshrathod.com/work/hr-analytics-agent/https://durgeshrathod.com/work/hr-analytics-agent/A natural-language analytics agent over sensitive workforce data, serving 2,000 concurrent users across isolated tenants. Reporting went from 2–3 days to under 30 seconds — but only after we stopped trusting the model with the boundary.Sun, 02 Aug 2026 00:00:00 GMTCase studytext-to-SQLmulti-tenant LLMconversational analyticsHR analytics AItenant isolationhttps://durgeshrathod.com/work/multi-agent-orchestration/https://durgeshrathod.com/work/multi-agent-orchestration/Two production agent systems on MCP and CrewAI — a project-management assistant and an autonomous lead-generation pipeline. What multi-agent buys you, what it costs, and the specific cases where a single agent with good tools wins.Sun, 02 Aug 2026 00:00:00 GMTCase studymulti-agent systemsModel Context ProtocolMCPCrewAIagent orchestrationhttps://durgeshrathod.com/work/telecom-throughput/https://durgeshrathod.com/work/telecom-throughput/A configuration-driven telecom data platform processing 200M configuration and 320M performance parameters per 15-minute cycle on Golang, Kafka, Kubernetes and PostgreSQL. The project that shaped how I think about bounded resources — and why I trust it more than any AI credential I have.Sun, 02 Aug 2026 00:00:00 GMTCase studyhigh throughput data pipelineGolang Kafka Kubernetesconfiguration-driven architecturetelecom performance managementdistributed systems