AI Engineering Essentials: Building Applications with LLM APIs
The developer's path from first API call to production: structured outputs, tool use, RAG, agents, evals, security, and deployment.
- 8 hours
- 8 modules · 24 lessons
- Certificate of completion
About this course
Calling an LLM API takes five lines of code. Building an AI feature that is accurate, safe, affordable, and maintainable takes engineering discipline. This course teaches the patterns that hold up in production.
Examples use Python and TypeScript with provider-agnostic patterns, and note where providers differ (Anthropic, OpenAI, Google, and open-weight models). Each module includes code you can adapt, and the course closes with a capstone: a grounded, tool-using assistant with an evaluation suite.
What's included
- In-depth technical lessons with annotated code samples
- Production readiness checklist for LLM features
- RAG design worksheet and eval suite starter template
- Prompt injection threat-model worksheet
- Capstone project brief
- Certificate of completion
- Lifetime access, including future updates to this course
What you’ll be able to do
- Choose models and providers based on capability, cost, latency, and data requirements
- Call LLM APIs robustly with streaming, retries, timeouts, and cost tracking
- Produce reliable structured outputs and implement tool use / function calling
- Build retrieval-augmented generation (RAG) systems with good chunking, retrieval, and citations
- Design agent loops with tools and MCP servers, with guardrails and budgets
- Build evaluation suites, defend against prompt injection, and ship AI features to production with monitoring
Syllabus
8 modules · 24 lessons · about 8 hours
01The LLM Application StackModels, providers, tokens, context windows, and the cost/latency/quality trade-offs that shape every design.3 lessons
- The LLM Application Stack Free preview18 min
- Choosing Models and Providers18 min
- Tokens, Context Windows, and Cost20 min
02Calling LLM APIs WellMessages and system prompts, sampling parameters, streaming, error handling, and retries.3 lessons
- Messages, System Prompts, and Conversation State20 min
- Sampling Parameters and Streaming20 min
- Errors, Retries, Timeouts, and Rate Limits22 min
03Structured Outputs & Tool UseJSON schemas, validation, function calling, and designing tools models can use correctly.3 lessons
- Structured Outputs with JSON Schema20 min
- Function Calling Fundamentals20 min
- Designing Tools Models Can Use20 min
04Retrieval-Augmented Generation (RAG)Embeddings, chunking, vector and hybrid search, reranking, and grounded answers with citations.3 lessons
- Embeddings and Chunking20 min
- Vector Search, Hybrid Search, and Reranking20 min
- Grounded Answers, Citations, and Retrieval Evaluation22 min
05Building AgentsAgent loops, tool design, the Model Context Protocol (MCP), planning, and safe autonomy.3 lessons
- The Agent Loop18 min
- Agent Tools, Planning, and Guardrails18 min
- The Model Context Protocol (MCP)22 min
06Evaluation & TestingEval datasets, code-based and model-graded evaluations, regression testing, and observability.3 lessons
- Eval Datasets and Code-Based Checks18 min
- Model-Graded Evals: LLM-as-Judge20 min
- Regression Testing in CI and Observability22 min
07Security, Safety & Cost ControlPrompt injection, data leakage, PII handling, abuse prevention, caching, and budgets.3 lessons
- Prompt Injection and the Lethal Trifecta22 min
- Data Protection, Output Handling, and Abuse Prevention18 min
- Caching and Cost Budgets20 min
08Shipping to ProductionArchitecture patterns, deployment, monitoring, and a capstone project.3 lessons
- Production Architecture Patterns20 min
- Monitoring, A/B Testing, and Safe Deployment18 min
- Capstone: A Grounded, Tool-Using Assistant24 min
Templates & resources included
Editable Word downloads after purchase, ready to adapt to your organization.
Full brief for the course capstone: scenario, functional, quality, and security requirements, six milestones with acceptance criteria, deliverables, and a weighted grading rubric for self-assessment or team review.
A ready-to-adapt evaluation suite: folder layout, a JSONL case format with sample cases, a pluggable Python harness with per-slice reporting and baseline gating, a judge rubric template, and a CI hookup.
A launch-gate checklist covering configuration, API resilience, outputs, tools, RAG, agents, evals, security, privacy, cost, observability, and deployment for any LLM-powered feature.
A step-by-step worksheet for mapping untrusted inputs, capabilities, and exfiltration channels in an LLM feature, assessing the lethal trifecta, and choosing layered controls and test cases.
A fill-in worksheet for designing a retrieval-augmented generation system: use case, corpus, access control, chunking, metadata, retrieval, reranking, grounded answers, evaluation plan, and operations.
Who this course is for
Prerequisites: Working knowledge of Python or JavaScript/TypeScript, HTTP APIs, and JSON.
Current, and backed by a guarantee
Updates are free for as long as you have access, and every purchase has a 14-day money-back guarantee.
Every certificate carries a credential ID that anyone can check at our verification page.
Frequently asked questions
How do I access my course after buying?
Right after checkout we email you a sign-in link. Click it and you're in. There's no password to remember. You can request a fresh link at any time from the sign-in page using the email you bought with.
How long do I have access?
Course and learning-path purchases include lifetime access, including future updates to those courses. The All-Access Pass gives you every course for as long as your yearly membership is active.
Is there a guarantee?
Yes. If a course isn't right for you, ask for a full refund within 14 days of purchase, as long as you've completed less than 30% of it.
Do I get a certificate?
Yes. When you complete every lesson in a course, you can download a certificate of completion with your name, the course, and the completion date.
Can my company pay with an invoice or buy for the whole team?
Yes. Choose 'For my team' on any course page to buy 3–250 seats with volume discounts. You'll get a Stripe invoice and can add your company tax ID. For 250+ seats, purchase orders, or custom terms, contact our team from the For Business page.