Generative AI and LLM Applications (GenAI): Prompting, RAG, Fine-tuning and Deployment

Learn generative AI live: how LLMs work, prompt engineering, the Claude, OpenAI and Gemini APIs, RAG with vector databases, LangChain, evaluation, fine-tuning, multimodal and deploying an LLM app.

Live batches with a trainer · Level: Intermediate · Duration: 40 hours · Language: English · Fee: ₹19,999

About this course

Generative AI is the fastest-growing skill in tech: companies are building assistants, document search, content and automation on large language models, and generative AI engineer is a new, well-paid role. This course teaches how LLMs work, how to build reliable applications on them and how to evaluate and ship them.

The course starts with how large language models work (tokens, transformers, training, context windows, costs) and prompt engineering that works in production (system prompts, few-shot, structured output, chain of thought), then the model APIs (Anthropic Claude, OpenAI, Google Gemini and open models with Ollama and Hugging Face) with streaming, tool use and structured JSON. You then build retrieval-augmented generation (RAG): document loading, chunking, embeddings, vector databases (Chroma, pgvector, Pinecone), hybrid search, reranking and citations, with LangChain and LlamaIndex.

You then make applications production-ready: evaluation (LLM-as-judge, test sets, RAGAS), guardrails, safety and privacy, caching and cost control, fine-tuning with LoRA and when it beats prompting, multimodal (images, audio, documents) and deployment with FastAPI, Streamlit and Docker. The course ends with two portfolio projects (a document assistant and an automation) and preparation for generative AI interviews. Every topic is built live, and every session is recorded.

What you will learn

  • Understand how large language models work: tokens, transformers, context windows, costs and limits
  • Write prompts that work in production: system prompts, few-shot, structured output and chain of thought
  • Use the Claude, OpenAI and Gemini APIs and open models (Ollama, Hugging Face) with streaming, tool use and JSON output
  • Build RAG applications: chunking, embeddings, vector databases, hybrid search, reranking and citations
  • Use LangChain and LlamaIndex, and know when plain Python is better
  • Evaluate LLM applications with test sets, LLM-as-judge and RAGAS, and add guardrails, safety and privacy
  • Fine-tune a model with LoRA and know when fine-tuning beats prompting
  • Build multimodal features with images, audio and documents
  • Deploy an LLM app with FastAPI, Streamlit and Docker, control cost, and finish two portfolio projects

Requirements

  • Python basics (functions, dictionaries, pip); no ML background needed
  • Best suited to developers, data analysts and scientists, product and automation professionals and graduates moving to AI roles
  • A laptop with a stable internet connection (8 GB RAM is enough; cloud notebooks are used for heavy work)

Syllabus

  • Module 1: How LLMs work (4 hours)
  • Module 2: Prompt engineering (5 hours)
  • Module 3: Model APIs (5 hours)
  • Module 4: Retrieval-augmented generation (8 hours)
  • Module 5: Evaluation, safety and reliability (5 hours)
  • Module 6: Fine-tuning and open models (4 hours)
  • Module 7: Multimodal and beyond text (3 hours)
  • Module 8: Deployment (3 hours)
  • Module 9: Projects and interviews (3 hours)

Upcoming live batches · Questions and answers · Book a free demo