Interactive AI engineering and Forward Deployed Engineer training
Learn how AI systems actually get built and shipped at customers.
Step through diagrams, break and fix real retrieval pipelines, size a model to a GPU, and practice customer conversations. Short chapters for software engineers who want to build with AI in the enterprise, whether or not "Forward Deployed Engineer" is on your next job title.
Try the free GPU memory calculator Or start with tokens and context windows
Start free: seven chapters, no signup
- LLM VRAM calculator: how much GPU memory a model needs, with the formulas shown.
- How LLM inference works: prefill, decode and the KV cache as an animation.
- Self-host an LLM or use an API? A decision tree and break-even calculator.
- RAG end to end: break and fix a retrieval pipeline in a lab.
- Embeddings and vector databases, explained on an interactive map.
- Tokens and context windows: a tokenizer and a context-budget lab.
- What a forward deployed engineer does, plus a customer-call simulation.
Or see the current forward deployed engineer jobs and posted salaries.
FDE-type roles are open right now
These are live postings pulled from company job boards. The pay is each company's own published base salary range, so you can see what the skills in this course are worth before you spend anything.
Company names and logos are trademarks of their owners, shown only to identify who is hiring. There is no affiliation or endorsement. Pay figures are base salary only (no equity or bonus), vary by level and location, and companies that do not publish pay show none. Always confirm details on the posting itself.
Why this course, and what it is worth
Those postings are not asking for more LeetCode. Reading them, the same asks keep showing up. This course teaches each one directly, as something you can practice rather than just read.
| What the postings ask for | Where you practice it |
|---|---|
| Turn a vague customer request into a scoped plan | Reframing a request, engagement gates, discovery and decomposition |
| Build with LLMs: RAG, tools, agents, structured output | The AI engineering foundations module: RAG, tool calling, MCP, agents, memory |
| Prove the AI works: evals, guardrails, security | Evals and eval frameworks, prompt injection and guardrails, security reviews |
| Run models inside the customer's environment | The inference engineering module: GPU memory sizing, how generation works, serving, quantization, scaling |
| Integrate with messy enterprise systems securely | OAuth and OIDC, APIs in the wild, customer networking, security reviews |
| Own it from POC to production, then hand it over | POC to production, demo craft, writing for executives |
Learn by operating the idea
Every chapter has something to click, break or tune, plus a scored quiz and a spaced-review queue so it sticks.
Built around the real job
Each chapter starts from a customer problem and ends with an "in the field" story, because the hard part is rarely the model.
Cheaper than one wrong turn
The subscription costs about what a few coffees cost per month. Cancel anytime. Try the free chapters first and judge the quality yourself.
Who this is for
Software engineers aiming for FDE roles
You can code. Postings also ask for customer discovery, scoping, security reviews, production ownership and applied AI. This course teaches those parts directly.
Engineers building with LLMs
You have called an API. You want to understand RAG, agents, tool calling, evals and guardrails well enough to make them reliable, not just demo-able.
Infrastructure and ML engineers
You want to understand what it takes to serve models: memory, batching, quantization, caching, autoscaling and cost.
Solutions engineers, consultants and technical PMs
You talk to customers and need technical depth that holds up in a room with their engineers and their security team.
Even if you never become an FDE
The skills are portable. Turning a vague request into a measurable problem, deciding between RAG and fine-tuning, proving an AI feature works, and shipping into a locked-down environment are what senior and staff engineers, startup founders and technical leads do every week. FDE is the sharpest place to practice them.
Why interactive diagrams teach better than reading
Most AI content is either a video you watch passively or a blog post you skim. Here, you operate the idea. Try it now: pick a question and see how a search by meaning chooses its results.
Predict, then reveal
You commit to a guess before the answer appears. Research on learning consistently finds that trying to recall or predict beats re-reading.
Break it, then fix it
Labs like the RAG retrieval lab and the GPU memory calculator let you change one thing and see which failure it causes or fixes.
Practice conversations
Branching customer scenarios show how a small wording choice changes trust and scope, with a debrief on every path.
Remember it later
Every quiz answer joins a spaced review queue, so what you learn on Monday is still there in a month.
What's inside
Chapters marked free open without signup. The rest unlock with a subscription. New chapters are added regularly and appear in your progress panel.
A course that stays current
AI tooling moves fast, and a static course goes stale in months. This one is maintained continuously: when something important changes, the affected chapter is updated and the progress panel on every lesson marks it Updated so you can catch up in minutes instead of rereading everything. Brand-new chapters appear with a New badge. A plain-language change log lists every update.
Fast-changing topics (tooling, model features, GPUs, protocols) are reviewed more often than the lasting ones (how to scope a problem, how to run a pilot). Your subscription includes every update and every new chapter while it is active.
Latest changes
Pricing
Monthly
Pay month to month. Cancel anytime from your account.
- Every chapter, lab and scenario
- Every update and new chapter
- Progress sync and spaced review
Annual Save $41 a year
Everything in monthly, billed once a year. Works out to about $6.60 a month.
- Every chapter, lab and scenario
- Every update and new chapter
- Updated-chapter alerts in your progress panel
- Share your login with a teammate if you like
Sign in with Google or an email link. Payments are handled by Stripe; we never see your card. Prices are in US dollars (USD); taxes are shown at checkout where they apply. Your progress is kept if you pause and come back. See the terms and privacy policy.
Questions
What do I need to know first?
Comfortable programming (Python or a similar language), basic git, and a rough idea of what an API is. No machine-learning background required.
Do I need to pay for an AI API key or a GPU?
No. Lessons run entirely in your browser with built-in simulations and calculators. Later hands-on labs let you plug in your own API key if you want to try real calls; the key stays in your browser and is never sent to us.
Is it affiliated with Palantir, OpenAI, Anthropic or any company mentioned?
No. Company names and logos appear only to identify employers in public job postings. All trademarks belong to their owners. The inference module follows the topic scope of the book Inference Engineering by Philip Kiely and links to it for further reading; the course text is independently written and is not endorsed by the author or Baseten.
How current is the information?
Job data is a dated snapshot from public job boards. Job-market statements cite public postings and published analyses, and each chapter notes where details can change. Anything fast-moving is flagged and reviewed more often.
Can I share my login?
Personal use is the intent, but we don't police sharing with a teammate or two. Please don't redistribute or republish the content. Paid content is delivered per user and carries an identifying mark.
Where is my progress stored?
On your device, and synced to your account once you sign in. You can also export and import it from the Review page.
Refunds?
Yes. Ask for a full refund within 24 hours of your first payment by emailing admin@fdeacademy.dev. See the terms for details. You can also cancel any time and keep access until the end of the period you paid for.
Who's behind this
FDE Academy is built by seasoned software developers who have worked for Fortune 500 companies in the United States and Canada. It is operated by 14328507 Canada Inc., Vancouver, British Columbia, Canada.
Questions, corrections or refund requests: admin@fdeacademy.dev. We read every message.