Phase 0 · Base Camp
MODULE: THE BRIDGE
Ship: Your first model-powered call
Learn: The LLM mental model, model APIs, async & streaming, cost and token thinking. No "Python from scratch" — you already code.
The map ends where the work begins. We're building in public.
Everyone is teaching you about AI. We're building real, production AI systems in public so you don't have to learn from toy projects. Not a course. An expedition with receipts.
Already got it? Read a dispatch
CHART · SHEET 1SCALE 1 : RAG → PROD
§ 01 — The terrain is moving
You learned the old ground well. The fear isn't that you're not good — it's that the map you trained on is being redrawn.
Backend and full-stack engineers, three to seven years in, watching the ground shift. The honest move isn't to panic or to chase hype. It's to walk the new terrain with someone who has already been ahead and plotted a route you can actually follow.
The Route
§ 02 — The plotted curriculum, station by station
A backend engineer gets a fast first win at base camp, then earns depth toward production. Every station ends with something shipped, deployed, and on your map.
Phase 1 · The Ascent
MODULE: APPLICATION ENGINEERING
Ship: A production RAG system
Learn: RAG done properly, agents & tool-calling, structured outputs, guardrails, and a first real deployment.
Phase 2 · High Ground
MODULE: PRODUCTION DEPTH
Ship: An eval & observability harness
Learn: Evaluation & observability, fine-tuning tradeoffs, serving & latency, reliability and failure modes — the moat the tutorials skip.
Summit · Capstone
MODULE: THE CROSSING
Ship: Your flag on the other side
Learn: Deploying, monitoring, and documenting one production-grade AI system. The centerpiece of a portfolio that is pure receipt.
§ 03 — Meet your guide
I'm building these systems because I needed to learn them. Now I'm logging the route.
I'm a software engineer transitioning into Applied AI. The biggest gap I found wasn't a lack of tutorials — it was a lack of production reality. Toy projects don't teach you how to handle rate limits, eval regressions, or agent loops that hallucinate and get stuck.
I don't claim to be an AI guru. I stay exactly one chapter ahead. I build a production system, log everything that breaks, and open-source the final architecture. That's what we study here. No fake expertise, just documented builds and honest receipts.
Logged in public
The Field Log
§ 04 — Every claim links to a real, logged discovery
We never teach what we haven't built. Each system below is being built in the open — we're logging the wins and the parts that break as we go. Tap any card to follow the repo as we build. Receipts, not badges.
DISCOVERY · 03 building
Eval & observability harness
Last update: 2 days ago
7 Experiments
3 Failures
2 Regressions caught
DISCOVERY · 02 building
Multi-tool agent runtime
Last update: 5 days ago
14 Tools integrated
99.2% Success rate
DISCOVERY · 01 building
RAG-over-Docs, re-ranked
Last update: 3 days ago
4.1s → 1.7s Latency reduced
+18% Retrieval accuracy
§ 05 — Field reports
Dispatches from
Dispatches from
the frontier
The owned channel. Build notes, what broke, and the route ahead — written from inside the work, sent to your inbox. Algorithm-proof.
Join the list Get the Roadmap.
Get the Roadmap.
Then the dispatches.
The best free route to becoming an Applied AI Engineer — plus a weekly field report from the build.
Build in public · no hype · unsubscribe anytime
§ 06 — Terms of the expedition
We guarantee the learning, not the job. Here's exactly what that means — in writing.
- Show up. ≥ 80% live-session attendance. The route only works if you walk it.
- Ship the work. All required projects submitted and deployed. Receipts, not attendance certificates.
- Go out into the field. Apply to your target roles. We prep you for the AI system-design interview.
- If you do all three and don't get across — re-enroll in the next expedition, free. The terms are public and checkable.
Founding Expedition · Cohort N° 001
Founding Expedition. Help shape the route.
Fifteen to twenty-five founding members. Live sessions, real project reviews, and a portfolio of deployed systems. Introductory pricing — it rises with every cohort's proof.
REFUNDABLE DEPOSIT · PAYMENT PLANS · EMPLOYER REIMBURSEMENT