LLM Data and Inference Engineering

This unit centers on preparing and managing data, selecting and testing large language models (LLMs), and establishing reproducible experiment and inference workflows. Students will learn ethical data practices, pragmatic engineering for dataset creation, experiment tracking, inference strategies, and fundamentals of fine-tuning to support downstream deployment and optimization.

Learning Objectives

• Analyze ethical, legal, and bias-related considerations in data collection, annotation, and use for LLM projects and document mitigation strategies
• Apply data engineering practices to build, clean, split, and version datasets with clear provenance for training and fine-tuning LLMs
• Demonstrate experiment management workflows for selecting LLMs, running controlled inference experiments, tracking metrics, and recording results reproducibly
• Evaluate LLM inference approaches and fine-tuning options by comparing trade-offs in latency, cost, accuracy, and scalability, and propose optimizations for production inference pipelines