The AI-Native Engineer — Founding Cohort

What's covered

Everything you'll wield by Friday.

Claude Code Agents Steering agents Codex MCP Copilot PRDs & plans OpenCode Tracer bullets GitHub Review discipline Pre-commit gates Token economics AFK agents Human-in-the-loop Context engineering Ship it Claude Code Agents Steering agents Codex MCP Copilot PRDs & plans OpenCode Tracer bullets GitHub Review discipline Pre-commit gates Token economics AFK agents Human-in-the-loop Context engineering Ship it

Who this is for

Built for working engineers — not beginners or vibe-coders

This is for you if…

✔You're a working engineer who already dabbles with Cursor, Copilot, or Claude — but can't yet hand an agent a real, messy production ticket and trust the output.
✔You can feel the field shifting, and you don't want to be the competent-but-not-AI-native engineer who gets left behind.
✔You're tired of demos on toy repos. You want to do this on your codebase, on real work.

This is not for you if…

✕You're brand new to coding, or looking to "vibe-code" an app without engineering fundamentals.
✕You want a passive video course to watch on 2× and never open your editor.
✕You're not willing to bring a real codebase or project to work in during the week.

What you'll walk out with

Proof you can do this — not a certificate that you watched videos

By Friday you won't have watched a course. You'll have shipped one real feature end-to-end with agents: planned it, steered it, reviewed it, and taken it to a PR you'd actually merge.

🚀

Ship one real feature end-to-end

Planned, steered, reviewed, and taken to a PR you'd actually merge.

🕹️

Direct an agent through production work

Real steering instead of babysitting autocomplete.

⚖️

Know when not to reach for an agent

The judgment that separates practitioners from hype.

🔍

Trust output you didn't write

Using the review discipline a regulated environment forces.

The 5 days

A live, hands-on week — you ship as you learn

One 45–60 minute live session per day, plus async work on your own codebase between sessions.

Day 1

Foundations: real speed vs. fake speed

Why "faster typing" is a trap, and what compressing the real idea-to-ship cycle looks like. You pick the feature you'll ship this week.

›Setup & environment (guided pre-work)
›Real speed vs. fake speed — the core model
›When not to hand work to an agent

Day 2

Steering an agent in a real codebase

How to actually make an agent behave — taught through the setup I run against a production system that moves money. You steer your agent on your feature.

›Context, steering files & skills
›The setup I run against a production money-movement system
›Steer your agent on your feature (live)

Day 3

Planning work agents can execute

Break a real production ticket into agent-sized phases — plus a straight look at what this actually costs.

›Break a real ticket into agent-sized phases
›PRDs, multi-phase plans & tracer bullets
›Token & cost economics — what this actually costs

Day 4

Trusting output you didn't write

The part everyone else skips — the review discipline that catches what agents get wrong. The difference between real speed and shipping a mess fast. This is the day that matters most.

›Feedback loops & pre-commit gates
›The review discipline that catches what agents get wrong
›Real speed vs. shipping a mess fast

Day 5

Ship it & what's next

Take your feature to a merged-PR-quality finish. Everyone ships. Then a map of the next level.

›Take your feature to a merged-PR finish (everyone ships)
›AFK agents & human-in-the-loop patterns
›What MCPs are and why they matter

Prerequisite: come with a real codebase or project. The "everyone ships" promise depends on it. Bring whatever you can legally and safely work in — a personal project works just as well.

Why me

Sal Shirgaleev

Self-taught engineer · ex-Dataminr · 60k on Instagram

I dropped out of school and taught myself to code. Ten years later, I've shipped production code at Dataminr and other AI companies — real-time AI systems and fintech platforms managing billions of dollars. Now I run a fleet of agents to ship every day, and share how with 60k people on Instagram.

I'm not a full-time course creator. I teach this because I do it every day — shipping real work with agents, not toy-repo demos. You're learning from someone in the trenches, not reading from a script.

Straight talk: this is the first cohort

No testimonials yet — I'm not going to fake them. That's exactly why it's $295 instead of $700, why founding members get direct access to me and help shape v2, and why the ship-it guarantee puts the risk on me, not you. You're trading being early for the lowest price this will ever be.

// founding trade: lower price + direct access + shape v2 ⇄ your feedback (and a testimonial if it lands)

Price & guarantee

The cheapest this will ever be

$295

Founding price · one-time · ~~next cohort $700~~

🛡️

The Ship-it guarantee Show up, do the work, and if you don't finish the cohort with a real feature you'd merge, you get a full refund. The risk is on me.

⏳ Enrollment closes Monday, July 27

–

Days

–

Hrs

–

Min

–

Sec

Claim your founding seat — $295

20 seats · enrollment closes Monday, July 27 · starts August 10

FAQ

Questions engineers actually ask

Do I need to be an AI expert already?

No. You need to be a working engineer comfortable in a codebase. If you can ship features by hand, you're ready.

What if I've never used Claude Code / agents seriously?

That's who this is for. You'll set up before Day 1 (guided) and go from casual dabbler to directing real work.

How much live time per day?

One 45–60 minute live session per day, plus async exercises you do on your own codebase between sessions.

Can I use my work codebase?

Bring whatever you can legally and safely work in. A personal project works just as well.

What if it's not for me?

The ship-it guarantee covers you — do the work and don't finish with a feature you'd merge, and you get a full refund.