Become a GenAI app developer: ship products on top of language models, including chat, search and RAG.

The mental model you need to build reliably.

1. Tokens, context and sampling
2. Model families and trade-offs
3. Pricing and latency
4. Streaming
5. Choosing a model

Make models return data your code can use.

1. Chat and response APIs
2. JSON schemas and structured output
3. Function and tool calling
4. Retries and fallbacks
5. Provider abstraction

Design what the model sees at each step.

1. System prompts that hold up
2. Few-shot examples
3. Dynamic context assembly
4. Long-context trade-offs
5. Prompt versioning

Answer from your own data accurately.

1. Chunking strategies
2. Embeddings and vector search
3. Hybrid search with BM25
4. Reranking
5. Citations and grounding

Fix the silent failures of basic RAG.

1. Query rewriting
2. Metadata and permission-aware filters
3. Multi-document and multi-hop questions
4. Freshness and re-indexing
5. When RAG isn't the answer

Let models act on systems and live data.

1. Tool design for LLMs
2. The Model Context Protocol
3. Building an MCP server
4. Tool-call budgets
5. Validating tool output

Know your app works, and keep it working.

1. Golden datasets
2. Retrieval metrics
3. Faithfulness and answer quality
4. LLM-as-judge
5. Evals in CI

Ship LLM apps that are fast, cheap and observable.

1. Caching and batching
2. Rate limits and queues
3. Tracing and observability
4. Cost monitoring
5. Deployment with Docker

Protect users, data and your system.

1. Prompt injection
2. Data leakage
3. Guardrails and moderation
4. PII handling
5. Abuse and rate limiting

Build and ship a complete RAG application.

1. Scoping a real use case
2. End-to-end build
3. Eval suite and dashboard
4. Deployment
5. Portfolio write-up