Module 1
LLMs as system components
Budgets, typed contracts, failure modes, and non-model fallbacks.
← All products
Self-paced course · Six chaptersInstant access
Software Engineer to AI Engineer
LLMs as system components, evals, knowledge strategies, agent loops, MCP, and production shipping — finish with an agentic feature design.
What it is
A self-paced course from software engineer to AI engineer: models as dependencies, evals, knowledge strategies, agent loops and MCP, production shipping, and a capstone agentic feature design — with HTML/CSS slides.
How it works
Pay once, open the library, work chapter by chapter on a feature you could actually ship, and finish with a design doc that includes evals, budgets, and guardrails.
Why it's worth it
You leave ready to implement — not with another chat demo that dies in a sprint review.
Built for
Practitioners shipping work, not collecting certificates.
Each chapter ends with exercises. The capstone is a deliverable you can present or implement.
Engineers adding LLM features to existing products.
Teams adopting MCP and tool-calling agent loops.
Tech leads setting quality bars for AI work.
Builders in regulated or high-stakes domains who need guardrails.
Curriculum
Six chapters. One agentic feature design.
Module 2
Prompting, tools, and evaluation
Golden tasks, adversarial cases, and traces that survive prompt churn.
Module 3
Knowledge strategies
RAG vs structured corpora vs fine-tuning — choose with an ADR.
Module 4
Agent loops and MCP
Tools, iteration limits, human gates, and encoding expertise as MCP.
Module 5
Shipping in production
Logging, cost/quality dashboards, and model upgrade drills.
Module 6
Capstone feature design
A complete agentic feature design ready to implement or pitch.
Chapters & slides
Visual decks for every chapter.
Preview the HTML/CSS teaching slides below. Buyers get the full decks in the library alongside written chapters.
Treat the model like a dependency
Latency, cost, hallucination risk, and version drift are production concerns — not prompt trivia.
- Inputs and outputs typed
- Budgets for tokens and time
- Fallback paths required
What you get
Chapters, slides, and a deliverable.
Six chapters from LLM dependencies through production ops.
HTML/CSS slide decks for every chapter.
Eval-set and agent-loop templates, plus a capstone design outline.
Lifetime library access — updates included. Download as a zip.
Frequently asked
About the course.
- When does it start?
- Whenever you do. Pay once, get instant email access, and start chapter one today. No cohort dates.
- How long does it take?
- Most people finish in four to six weeks at one chapter per week — roughly 3–5 hours per week including exercises and the capstone.
- Is it live or self-paced?
- Self-paced. Written chapters plus HTML/CSS slide decks — no live sessions required.
- Who is this for?
- Software engineers comfortable shipping backend or full-stack features who want a production-minded path into LLMs, tools, and agents.
- What does it cost?
- One-time purchase — see the pricing page for the current USD price. 14-day refund if you have not used the library.
Get access
Pay once. Start chapter one today.
After checkout you receive an email with a library link and access token. Work through chapters and slides at your own pace.