LLM Engineering: Prompting, RAG, Fine-Tuning, and RLHF

Course Overview

This course provides practical engineering skills to build production-ready Large Language Model (LLM) systems. It covers advanced prompt engineering, retrieval-augmented generation (RAG) pipelines, parameter-efficient fine-tuning methods like LoRA, and reinforcement learning from human feedback (RLHF) alignment techniques. The course bridges theoretical foundations with hands-on labs to enable learners to design, build, and align specialized LLM systems.

Key Takeaways

• Design advanced prompts including zero-shot, few-shot, and chain-of-thought reasoning for reliable, high-quality outputs.
• Architect and build complete RAG systems covering ingestion, retrieval, synthesis, and rigorous evaluation.
• Fine-tune base models into specialized assistants using parameter-efficient fine-tuning methods such as LoRA, QLoRA, rsLoRA, and LoRA-FA.
• Align LLM behavior to human preferences using the full RLHF workflow and algorithms including PPO, DPO, and GRPO.
• Combine prompting, retrieval, fine-tuning, and alignment into end-to-end, production-ready LLM solutions.

Prerequisites

• Understanding of Transformers and LLMs including neural networks, attention mechanisms, and LLM training and generation processes.
• Intermediate Python skills with comfort reading and writing functions, loops, and basic data structures.
• Basic machine learning intuition including vectors, embeddings, and training concepts like loss and gradients.

Target Learners

• Software engineers and developers aiming to move beyond API usage to architect reliable, custom LLM systems.
• Data scientists and machine learning practitioners seeking practical skills in RAG, fine-tuning, and alignment.
• Computer science students and technology professionals desiring a deep, principled understanding of LLM engineering.