The Blog

Notes on AI & Engineering

Writing about RAG pipelines, LLM agents, and what it takes to bridge Generative AI with production-grade full-stack systems.

AITelegramClaude

How I Built a Personal AI Assistant on Telegram Using Claude

A step-by-step guide to connecting Anthropic's Claude AI to Telegram, including real use cases, hidden risks, and everything it costs.

Aug 10, 2026 · 7 min read

RAGLangChainFAISS

Building Production-Ready RAG Pipelines: Lessons from DocuMind

Retrieval-Augmented Generation looks simple in a tutorial. Shipping it with sub-2-second latency, conversational memory, and predictable costs is a different problem entirely.

Jun 12, 2026 · 2 min read

AgentsLangGraphLangChain

From ReAct to LangGraph: Designing Reliable LLM Agents

A plain ReAct loop is enough to impress in a demo. It falls apart the moment you need observability, conditional branches, or a human in the loop. Here's why I rebuilt AgentFlow on LangGraph.

May 3, 2026 · 2 min read

CareerGenAIFull-Stack

Bridging Full-Stack Engineering and Generative AI

The most valuable AI engineers I know aren't the ones who only understand transformers. They're the ones who can also ship the dashboard, the auth flow, and the deployment pipeline around the model.

Mar 21, 2026 · 2 min read