Course06 · LLMs and Generative AI

1. Day 239 How Large Language Models Are Trained planned
2. Day 240 Pretraining, Fine-Tuning, and RLHF planned
3. Day 241 The Model Landscape: Claude, GPT, Gemini, Llama planned
4. Day 242 Open Weights versus Closed APIs planned
5. Day 243 Tokens, Context Windows, and Sampling planned
6. Day 244 Capabilities, Limits, and Hallucination planned
7. Day 245 Benchmarking Models Yourself planned

Project: Model Comparison Study — Design a ten-task benchmark and compare two accessible models on it, documenting failure patterns and cost.

1. Day 246 Prompting Fundamentals planned
2. Day 247 System Prompts and Role Design planned
3. Day 248 Few-Shot Examples and Chain of Thought planned
4. Day 249 Structured Output: Getting Reliable JSON planned
5. Day 250 Prompt Patterns and Templates planned
6. Day 251 Prompt Injection and Safe Prompting planned
7. Day 252 A Tested Prompt Library planned

Project: Reusable Prompt Library — Build a tested library of parameterized prompts for five recurring tasks, each with evaluation examples.

1. Day 253 First Calls to the Claude API planned
2. Day 254 The OpenAI-Compatible Ecosystem planned
3. Day 255 Streaming Responses planned
4. Day 256 Tool Use and Function Calling planned
5. Day 257 Working with Images and Documents planned
6. Day 258 Cost, Caching, and Rate Limits planned
7. Day 259 Building a CLI Assistant planned

Project: Command-Line AI Assistant — Build a streaming CLI assistant with conversation memory, tool use, and a cost meter.

1. Day 260 What Embeddings Are planned
2. Day 261 Semantic Similarity Search planned
3. Day 262 Vector Databases planned
4. Day 263 Chunking Strategies planned
5. Day 264 Hybrid Search and Rerankers planned
6. Day 265 Evaluating Retrieval Quality planned
7. Day 266 Semantic Search over Your Own Notes planned

Project: Semantic Search Engine — Build semantic search over your own notes with embeddings, a vector store, and a retrieval-quality evaluation.

1. Day 267 The RAG Architecture planned
2. Day 268 A Minimal RAG System from Scratch planned
3. Day 269 RAG over PDFs and Messy Documents planned
4. Day 270 Citations and Grounded Answers planned
5. Day 271 Advanced RAG Patterns planned
6. Day 272 Evaluating RAG Systems planned
7. Day 273 A Documentation Assistant planned

Project: Documentation Q&A Bot — Build a RAG assistant over a real documentation set with citations and a measured answer-quality eval.

1. Day 274 Prompting versus RAG versus Fine-Tuning planned
2. Day 275 Fine-Tuning with LoRA planned
3. Day 276 Building Fine-Tuning Datasets planned
4. Day 277 Running Local Models with Ollama planned
5. Day 278 Quantized Inference and llama.cpp planned
6. Day 279 Serving Open Models planned
7. Day 280 Fine-Tune and Serve Your Own Model planned

Project: Local Fine-Tuned Model — Fine-tune a small open model with LoRA on a task dataset and serve it locally, comparing before and after.

1. Day 281 How Diffusion Models Generate Images planned
2. Day 282 Image Generation in Practice planned
3. Day 283 Speech: Recognition and Synthesis planned
4. Day 284 Video and Music Generation planned
5. Day 285 Multimodal Models planned
6. Day 286 Generative AI Ethics and Copyright planned
7. Day 287 Section Project: A Multimodal Application planned

Project: Section Project: Multimodal App — Build an application combining at least two modalities — for example, speech in, grounded text out, generated image alongside.