LLM Data Selection and Experimentation

This unit guides students through ethical dataset creation, practical engineering for prototyping, and the practical choices involved in selecting and using large language models. Students will learn experiment management, inference optimization techniques, and basic fine-tuning workflows to prepare models for downstream units on deployment and production integration.

Learning Objectives

• Analyze ethical, legal, and societal implications of data collection and labeling for LLMs, and identify bias mitigation and data governance strategies
• Apply practical engineering practices to build, document, and version datasets and lightweight pipelines that support reproducible LLM experiments
• Evaluate and select an appropriate LLM and inference strategy by comparing model capabilities, cost, latency, privacy, and task-specific performance trade-offs
• Demonstrate experiment management, inference optimization, and fine-tuning workflows by designing, executing, and documenting experiments that track metrics, reproducibility, and deployment-ready optimizations