AI-101

AI Engineering Foundations

The vocabulary, mental models, and base-level systems work every engineer working with LLMs is expected to know in 2026.

ABOUT THIS COURSE

What you will learn.

AI engineering, as a discipline, has stabilized faster than most people realize. The patterns that work — and the patterns that quietly fail in production — are now well-known among the small group of engineers building these systems for a living. The problem is that almost none of that knowledge is written down.

This course is the canonical introduction. We start from the bottom: what an LLM actually is as a piece of software, what its API contract really guarantees, how tokens and context windows behave under load. From there we work upward through prompt structure, tool use, streaming, structured outputs, function-calling failure modes, retry strategy, cost accounting, and the basics of evaluation. By the end you will have built four small but realistic systems, broken them on purpose, and understood why they broke.

This is the prerequisite course we ask other Solutech students to take if they show up unable to explain why their first RAG system is bad. It is not a survey; it is the engineering foundation.

WHAT YOU’LL BUILD

Four substantial projects.

Project 01

A correctness-checked summarizer

Build a summarization endpoint that catches three classes of hallucination before responding to the caller.

Project 02

A streaming chat with cancellation

Implement streaming end-to-end with mid-flight cancellation, retry budgets, and a real cost ceiling per request.

Project 03

Structured output with a self-correcting parser

Return validated JSON from an unreliable model, including a recovery loop that does not run forever.

Project 04

Tool-use with adversarial inputs

Wire up a tool-using assistant against a small adversarial test set and ship it past a code review.

CURRICULUM

Week by week.

FIT

Who this is for — and who it is not.

For you if

Engineers with one or two years of professional experience who have shipped a feature against an LLM API and want to actually understand what they did.
Senior engineers from adjacent disciplines (backend, mobile, data) joining an AI team.
Technical PMs and EMs who want to be useful in design reviews involving model behavior.

Probably not for you if

Engineers who have already shipped two or more production LLM systems and run their own eval suite.
People looking for a survey of every model on the market — we are opinionated about which APIs we use.
Researchers who want to study training dynamics — this is engineering, not modeling.

YOUR INSTRUCTOR

Taught by an operator.

Staff Engineer

Marcus Hale

Marcus spent eight years at Stripe, where he led the team responsible for the Radar risk-decisioning platform. Before that he wrote infrastructure at Square and Twilio. He thinks about LLM systems the way payments engineers think about payments — with a healthy paranoia about retries, idempotency, observability, and the long tail of failure modes that only show up at 3 a.m. on a Saturday.

FAQ

Questions we’re asked often.

AI-101 · Next cohort starts soon