Academy · v1 · 2026
From zero to shipping
AI in production.
A practical academy for engineers. Six modules, dozens of lessons, capstone projects. Build agents, RAG systems, and production AI services on your own machine.
Trusted by engineers from
800+ engineers enrolled.
$ docker compose up vllm postgres
→ vLLM ready on :8000
→ pgvector extension loaded
$ python -m agent.run
→ tool: search_docs() ✓
→ tool: write_to_index() ✓
→ agent ready · 17ms p50
Architecture · the stack you ship
The problem · why engineers get stuck
Most AI tutorials stop right
before production starts.
04 · observed
today- 01Cloud API bills creeping into the thousands per month.
- 02Prototype demos that crumble the moment real traffic hits.
- 03No reproducible eval — "works on my laptop" debugging.
- 04Agents that hallucinate tools and crash silently.
04 · shipped
after academy- 01Local-first inference, predictable cost, full data control.
- 02Containerized services with measurable latency budgets.
- 03A real eval harness, RAG triad, regression tests.
- 04Agents that recover from failure and log every step.
"The eval harness alone paid for the academy. We caught regressions that would have shipped to prod."
Audience · two clear signals
Built for you. Or not.
Built for you
- You write code daily and want a real production AI stack.
- You can read documentation and run things locally.
- You prefer first-principles to copy-paste tutorials.
- You want to ship, not just play with demos.
Look elsewhere
- You expect hand-holding through every keystroke.
- You only want pre-built no-code tools.
- You are not interested in evaluating or testing models.
- You only want managed cloud APIs and never local infra.
Curriculum · six modules
What you will build.
Each module ends with a capstone you can put in your portfolio.
Instructor
Hands-on, opinionated, in production.
Course author with a decade of ML and AI engineering experience. Tutorials are written while shipping the same systems in real products — what works, what breaks, what to skip.
- Years building ML
- 10+
- Engineers enrolled
- 800+
- Users in production
- 100M+
FAQ · 10 questions · common objections
Common questions.
Final call · the last step
Stop reading tutorials.
Start shipping AI.
- 01Six structured modules · capstone projects
- 02Local-first stack · runs on your laptop
- 03Production patterns · evals · observability
- 04Discord community · monthly office hours
"Local-first stack changed how I prototype. No more API bills, no more rate limits during dev."
Pricing · two ways
One academy. Two ways to access.
Lifetime access · all updates · community
- All current and future modules
- Capstone projects with reference code
- Quizzes with auto-graded feedback
- Discord community
- Invoiceable for teams
- 30-day refund
Stripe · SSL · invoice on request