What I Learned About How LLMs Work — A Developer's Summary
After watching Andrej Karpathy's Intro to LLMs, here's my plain-English breakdown of how Large Language Models are built, trained, and why they think the way they do.
🚀 Full Stack Dev to AI Explorer
Hi, I'm Ravi Kishore Thella
9 years of building production apps with JavaScript, React, Python, Java, Spring Boot. Now exploring AI agents, LLMs, RAG, and LangChain. Learning in public.
Tech Stack
JavaSpring BootReactJavaScriptTypeScriptLangChain.jsSpring AILangGraphRAGOpenAISupabaseGCPAWSDockerPostgreSQL
Recent Posts
View allUnderstanding LangChain — What It Is, Why It Exists, and When to Use It
A developer's guide to LangChain — the problem it solves, how it works, its real advantages, and the honest disadvantages nobody talks about.
05/01/2026·8 min read
RAG From Scratch — How to Make an LLM Answer From Your Own Documents
I watched AI Jason's RAG from Scratch course so you don't have to wade through 2.5 hours. Here's the core of what RAG is, how it works, and why it matters — explained the way I wish someone had explained it to me.
Featured Projects
View allRAG Chatbot — Java Interview Coach AI
✅ CompletedA Retrieval-Augmented Generation chatbot that answers Java interview questions from personal markdown notes. Built with Next.js 16, FastAPI, LangChain, Groq (Llama 3.1), HuggingFace Endpoint Embeddings, and Neon Postgres with pgvector. Frontend and Python backend co-deployed as a Vercel monorepo.
RAGLangChainGroqFastAPINext.jsNeonpgvectorHuggingFace