Pathway 7: "I'm a Data Scientist Adding LLMs to My Toolkit" (Data Scientist / Analyst)

Target audience: Data scientists and analysts who already know Python, pandas, scikit-learn, and basic deep learning

Goal: Learn where LLMs complement (not replace) your existing ML models, how to build hybrid pipelines, and how to use LLMs for feature engineering, classification, extraction, and analytics.

Chapter Guide

• Skim Ch 00: ML and PyTorch Foundations (refresh if needed) refresh PyTorch if needed
• Skim Ch 01: NLP and Text Representation (refresh if needed) refresh NLP basics if needed
• Skim Ch 02: Tokenization (refresh if needed) refresh tokenization if needed
• Focus Ch 03: Sequence Models and Attention bridge from classical sequences to attention
• Skim Ch 04: The Transformer Architecture (unless curious about internals) optional deep dive into transformer internals
• Focus Ch 05: Decoding and Text Generation understand how generation differs from prediction
• Skim Ch 06: Pre-training and Scaling Laws (especially Section 06.4 on data curation) especially data curation for LLM training
• Skim Ch 07: The Modern LLM Landscape choose the right model for each analytical task
• Focus Ch 10: Working with LLM APIs call LLMs from your data pipelines
• Focus Ch 11: Prompt Engineering prompt LLMs for extraction and classification
• Focus Ch 12: Hybrid ML+LLM Architectures combine scikit-learn and LLMs in one pipeline
• Skim Ch 13: Synthetic Data Generation generate labeled data when labels are scarce
• Skim Ch 14: Fine-Tuning Fundamentals (especially Section 14.6 on classification) fine-tune for classification and extraction tasks
• Focus Ch 19: Embeddings and Vector Databases embed your datasets for similarity search
• Focus Ch 20: RAG build retrieval over your knowledge bases
• Focus Ch 28: LLM Applications applied patterns for analytics and reporting
• Focus Ch 29: Evaluation and Experiment Design evaluate LLM outputs with your existing metrics
• Skim Ch 34: Emerging Architectures scaling laws and new model designs to watch
• Optional Ch 35: AI and Society responsible AI context for data practitioners