Large language models (LLMs) now power search, customer support, coding assistants, and enterprise knowledge tools, and building on them reliably has become a core developer skill. This Specialization takes you from LLM fundamentals and prompt engineering to tool-calling applications, retrieval-augmented generation (RAG), fine-tuning, and LLMOps. Every course is built around follow-along demos in Python, so you write and run working code as you learn.
By the end of this Specialization, you will be able to:
Explain how transformers, tokenization, and attention shape LLM behavior.
Design structured, reasoning, and injection-resistant prompts for real tasks.
Build LLM applications with provider APIs, tool calling, LangChain, and LlamaIndex.
Implement RAG pipelines using embeddings and vector databases.
Evaluate retrieval quality and fine-tune models with LoRA and QLoRA.
Deploy and monitor GenAI services with FastAPI, Docker, and LLMOps practices.
This Specialization suits software developers, Python programmers, data scientists, and ML engineers who want to add generative AI features to their products. Working knowledge of Python and APIs will help you get started, and no prior machine learning experience is required.
Enroll now to start building LLM applications that are grounded, measurable, and ready to scale.
Projet d'apprentissage appliqué
Across the Specialization, you will build a progressively more capable LLM system. In Course 1, you will assemble an end-to-end prompted application that combines structured prompts, reasoning techniques, prompt injection defenses, and context management within a token budget. In Course 2, you will build a tool-using, retrieval-backed LLM application that calls Python functions and answers questions from your own documents using LangChain, LlamaIndex, and a vector database. In Course 3, you will complete a capstone that turns a GenAI application into a deployed, guarded, and monitored service: improving retrieval with re-ranking and hybrid search, scoring it with RAGAS, serving it through FastAPI and Docker, and tracing its behavior in production. Each project mirrors the workflow developers follow when shipping LLM features at work.
