This course is designed for individuals who wish to build real artificial intelligence systems hands-on, rather than just studying theory. Covering everything from language models to multimodal agents, you will follow the complete journey of an AI engineer by creating practical projects at each learning stage.
What to Expect:
-
Project 1: LLM Playground
Create your own sandbox for working with language models. Learn the basics of tokenization, architectures (GPT, Llama), text generation methods, post-training techniques (SFT, RLHF), and quality assessment metrics. -
Project 2: Customer Support Chatbot with RAG and Prompt Engineering
Practice model adaptation through fine-tuning and PEFT. Learn prompt engineering techniques (few-shot, zero-shot, chain-of-thought) and explore Retrieval-Augmented Generation, involving search, indexing, and quality evaluation. -
Project 3: “Ask-the-Web” Agent
Build an agent that interacts with tools and the web, focusing on agent systems, planning, reflection, and multi-agent workflows. Learn about tool calling and methods to evaluate agent performance. -
Project 4: Deep Research with Reasoning Models
Work with advanced reasoning LLMs (e.g., OpenAI o1, DeepSeek-R1). Explore inference methods like Chain of Thought and Tree of Thoughts, and learn to train on reasoning data using SFT, reinforcement learning with verifier, and self-refinement. -
Project 5: Multimodal Agent (Text - Image/Video)
Generate images and videos using techniques like diffusion, GAN, and VAE. Understand the architectures and training of diffusion models and build an end-to-end T2I and T2V system with evaluations of generation quality.
This course provides hands-on experience and comprehensive knowledge of AI engineering through diverse projects.