
Already know some of this? A 2-minute test skips the lessons you know.
Take the placement testA free 21-lesson course built from Sheryians AI School's 3h 52m of YouTube video. 3 days at 90 minutes a day, with study notes and a quiz in every lesson. A 2-minute placement test skips what you already know.
Taught by Sheryians AI School on YouTube. Videos play in YouTube's own player, so the creator keeps every view.
Understand the course roadmap and identify the five prerequisites (Python, ML/DL basics, Transformers, NLP) needed before starting Generative AI development.
Define Large Language Models, describe their training on massive text corpora using GPUs, and distinguish LLMs from search engines and human reasoning.
Differentiate between chat models (text generation), embedding models (vector representation for search/RAG), and multimodal models (image/audio processing) with provider examples.
Classify prompt types (simple, system+user, template, structured), explain role-based prompting (system/user/assistant), and describe reusable prompt templates.
Explain LLM chains for multi-step tasks and memory concepts (short-term vs long-term) for maintaining conversation context in LangChain.
Create a Python virtual environment, obtain API keys from OpenAI, Google, Groq, Mistral, and Hugging Face, and manage dependencies with a version-locked requirements.txt using uv.
Use LangChain's `init_chat_model` to connect to OpenAI, Google Gemini, Groq, and Mistral models with a consistent interface, and handle provider-specific model classes.
Download and run smaller Hugging Face models (e.g., TinyLlama, DeepSeek) locally using `ChatHuggingFace` pipeline, avoiding API costs and enabling offline inference.
Generate vector embeddings for documents and queries using OpenAI and free Hugging Face embedding models, and configure dimensions for storage efficiency.
Implement a functional chatbot that maintains chat history using message lists, enabling context-aware multi-turn conversations with role distinction.
Apply System, Human, and AI message roles to structure conversations, solve raw-string ambiguity, and prepare for LangChain's message history management.
Wrap the chatbot logic in a Streamlit web UI with streaming responses, role selection (Funny/Sad/Angry), and session state memory persistence.
Design a prompt template for a movie metadata extraction task (CineSage), defining system role, human placeholder, and structured output rules for consistent AI behavior.
Define a Pydantic schema (MovieData) with typed fields, create a PydanticOutputParser, and integrate it into the prompt template to enforce valid JSON output for database storage.