Application-based · 12 weeks
Learn AI engineering in 12 weeks? You can if you ship
The schr.ai internship is a seat at the bench of a working AI practice — the AI practice of Schraff, a family business in its third generation since 1976. You ship agents, automations, and knowledge systems that run for real users, with weekly build sessions, code review, and direct mentorship the whole way. Not exercises. Production.
Small cohort. Applications are read as they arrive, by a person.
1// Week 6: your PR, reviewed line by line
2feat(intake): route edge cases to humans
3
4 + agent/router.ts +84
5 + agent/router.test.ts +112
6 ~ pipeline/intake.ts ±23
7
8// review: "Good catch on the retry loop.
9// Now make the failure path this clean."How it works
/// 01
1:1 onboarding call
We start with a real conversation: where you are, what you can already build, and what you want to be able to build. You leave day one with a 12-week plan with your name on it.
/// 02
Weekly build sessions
Live, every week: demos of what shipped, code review in the open, hot seats on whatever has you stuck, and planning the next build. Bring work. Watching doesn't count.
/// 03
Milestone check-ins
1:1 reviews at weeks 4, 8, and 12. An honest read on where you stand, what to double down on, and where the plan needs to change. The plan serves you, not the other way around.
/// 04
Production, not coursework
Nothing you build here is a toy. The practice runs a production fleet for real client brands — retailers, schools, service businesses — and your work ships into it. It gets monitored, it breaks, it gets fixed. That's not the downside. That's the education.
Deploy log · example
The 12-week roadmap
Day 1
1:1 onboarding call
Goal setting, skills assessment, and your personal 12-week plan.
personalized strategy sessionWeeks 1–4
Foundation
Learn the practice's toolchain, ship your first automations, and put your first reviewed pull requests into a real codebase.
4 build sessions + async supportWeek 4
First check-in
1:1 review of what you've shipped and how you work. Adjust the plan.
1:1 milestone reviewWeeks 5–8
Build
Take ownership of a system end-to-end (an agent, a pipeline, a knowledge base) from design through deploy.
4 build sessions + async supportWeek 8
Second check-in
1:1 deep-dive on your project: what's working, what isn't, and what production is teaching you.
1:1 optimization sessionWeeks 9–12
Ship
Harden your system for real users: monitoring, failure handling, documentation, and the unglamorous work that makes software trustworthy.
4 build sessions + async supportWeek 12
Final review
A 1:1 walkthrough of everything you built, what it says about you as an engineer, and what comes next.
portfolio review + next stepsWhat you get
1:1 onboarding + milestone check-ins
Personal reviews at day 1 and weeks 4, 8, and 12: your plan, adjusted to reality.
Weekly live build sessions
Demos, open code review, hot seats, and planning with the cohort, every week.
Direct mentorship from working engineers
Your reviewers build and run these systems for paying clients. You get their real standards. There is no curriculum version of that.
Real projects in a real codebase
Agents, automations, and RAG pipelines that serve actual users, with the access and guardrails to work on them safely.
Line-by-line code review
Every pull request reviewed the way production code gets reviewed. This is where most of the learning happens. It stings. Then it sticks.
A portfolio of shipped systems
You leave with running software you built, deployed, and can walk anyone through. Not certificates. Evidence.
What interns actually build
The same categories of systems the practice ships for clients. These are the shapes of project you'll take on: scoped to you, supervised throughout.
An intake or support agent
An autonomous agent that reads, qualifies, drafts, and routes, with the judgment calls escalated to humans and every action logged.
Example trace
new_inquiry → qualify ✓ → draft ✓edge case → escalate to human
A workflow automation
A pipeline that moves data between the systems a business already runs: triggered, transformed, retried, and observable.
Example run
webhook → transform → crm ✓retry(1) on timeout → delivered ✓
A RAG knowledge system
A question-answering system over real documents: chunking, embeddings, retrieval, and answers that cite their sources. The practice's own brain indexes 843,000+ memories the same way.
Example answer
Q: which plan includes SSO?A: Team and up — cites pricing.pdf §2
What the program asks of you
The schr.ai Internship
12 weeks inside a working AI practice.
- 10–15 focused hours a week, including the weekly build session.
- Show up and ship: attendance at sessions and check-ins, work in the open.
- Take review seriously: the feedback is direct because the systems are real.
Program terms, including structure and expectations, are covered in detail on the screening call.
Frequently asked questions
01Who is this program for?
Builders who are serious about learning AI engineering by doing it: students, career-switchers, self-taught developers, or people already in tech who want production experience with AI systems. What matters is that you'll show up and ship for 12 weeks. That's the whole filter.
02How much time does it take each week?
Plan for 10–15 focused hours per week: the weekly build session, your project work, and responding to code review. The program is designed to fit alongside a job or studies, but it does not work if it's squeezed into Sunday nights.
03Do I need to be a strong programmer already?
You need to be able to write and run code — any language, any level of polish. You do not need AI experience; the toolchain, patterns, and review process are what the program teaches. The screening call is where we figure out honestly whether the gap is crossable in 12 weeks.
04Is the internship paid? What does it cost?
Program terms, including compensation and cost, are discussed one-on-one at the screening call. There is nothing to buy on this page, and applying costs nothing.
05What happens after the 12 weeks?
You leave with running systems you built, a review history that shows how you work, and a final 1:1 focused on what's next. Where there's mutual fit, standout interns are the first people the practice looks to when work grows, but that's earned, not promised.
06How is this different from a course or bootcamp?
Courses sell information; this is an apprenticeship. You work in a real codebase, on systems with real users, held to the same review standard as the practice's own work. There is no curriculum to complete; there is software to ship, and mentors making sure you ship it well.
Ready to start?
Apply below. If it looks like a fit, we'll set up a screening call and take it from there.
Apply below
A short application so we understand your goals and where you're starting from.
Screening call
A 30-minute call to make sure the fit is real, and to cover the program terms in full.
Onboarding & go
Your 1:1 onboarding call kicks off day one. Build sessions start the same week.