All Posts

RAG solves two fundamental LLM limitations: hallucination and knowledge cutoff. A complete hands-on guide covering the full pipeline — document loading, text splitting, HuggingFace embeddings, InMemoryVectorStore, retriever, and a full LCEL RAG chain. Free, no API cost needed.

How to chain LangChain components with the pipe operator | to build clean, modular, streamable pipelines. Covers the Runnable interface (invoke, batch, stream), RunnablePassthrough, RunnableParallel, RunnableLambda, and two real use cases: word count chain and subject-based routing.

The 4 fundamental concepts you need to master before anything else in LangChain: Models (invoke, stream, batch), Messages (System, Human, AI), Prompt Templates, and Output Parsers (StrOutputParser, with_structured_output). Built on top of OpenRouter — free tier, no credit card needed.

You built a model. It scores 80% accuracy. You're happy. But your model is broken , and you don't know it yet. A complete end-to-end ML project on customer churn: imbalanced data, Chi-squared feature selection, feature engineering, sklearn Pipeline, Recall vs Accuracy, GridSearchCV, and ROC curve.

a brief intro to ML through titanic dataset