Become an LLM Engineer · AthlyticZ

Career Programs

Become an LLM Engineer

Data scientists who want to build, evaluate, and ship LLM-powered tools rather than prompt them.

Opens later16 weeks · Foundations to Retrieval and evaluation to Capstone and Film Room
Is this you

Three things people say before they start.

If none of these are familiar, this is probably not the right program, and the call will tell you that faster than the page will.

  • You have wired up an API call and you do not know whether the output is any good.
  • You can demo it. You could not defend it.
  • Somebody wants a number for whether it is working, and you do not have one.
The IDE, compiling a Stan programA Stan model compiling on baseball data inside the provisioned environment. This is the editor students actually work in.

This is a real session, not a promotional cut. LLM Tools runs live alongside the program.

The game plan

16 weeks, week by week.

Not a reading list. Each phase carries the skills you work on, what exists at the end of it, and what you can do by then.

Languages and tools across the program
PythonR
Weeks 1 to 4

Foundations

The data and modeling layer underneath the part everyone talks about.

In these weeks
  • Working fluency in Python and R
  • Reproducible pipelines and project structure
  • End-to-end git workflow: branches, review, and QA gates before anything lands
  • Mentor one-to-one every second week, on your own work
What exists at the end
A reproducible pipeline
By now you can

Build a data pipeline that runs the same way twice.

Weeks 5 to 10

Retrieval and evaluation

Where this stops being a demo: measuring what the system does instead of asserting it.

In these weeks
  • Retrieval pipelines built to be inspected
  • Evaluation harnesses, and the eval set that makes them mean something
  • Model validation gates before anything ships
  • Simulated technical interviews, recorded and reviewed
  • Mentor one-to-one every second week, on your own work
What exists at the end
A retrieval pipelineAn evaluation harness
By now you can

Say how well your system works and show the number that says so.

Weeks 11 to 16

Capstone and Film Room

The build, the orchestration around it, and the rooms it has to survive.

In these weeks
  • Capstone tool with its evaluation harness
  • Deployed and monitored, not just running locally
  • AI orchestration training: planning work for delegation, delegated builds, browser QA
  • Front-office presentation practice: the recommendation, and the pushback
  • Simulated technical interviews, recorded and reviewed
  • Mentor one-to-one every week through the capstone
What exists at the end
The capstone toolAn eval reportA portfolio writeup
By now you can

Explain to a non-technical room what your system can and cannot be trusted to do.

Dates are not set yet. The waitlist is where they are announced, and it is one message when they land.

Join the waitlist
The Coaching Room

Practice with somebody on the other side of the table.

AI coaching is built into the practice sessions. The coach takes the stakeholder's part, and you work the problem in whatever medium the question actually calls for.

What that means literally: AI coaching built into practice sessions, stakeholder simulation where you present to a general manager and defend a decision to a director, and no-code visual builds alongside code builds, because plenty of the work is a sketch or a plan rather than a script.

AI as copilot, not autopilot. We teach you to direct the machine and defend the output. The part that is scarce is not producing an answer; it is knowing whether the answer is right and being able to say why in a room that will push back.

AthlyticZ Worlds

Practice worlds, built from the course itself.

Between the live sessions and the mentor calls, this is where the repetition happens. Not a quiz: the world is generated from the course's own code and data.

AthlyticZ WorldsA practice world worked through end to end.

Interactive practice worlds built from each course's own code, data and recordings. Every quantity computed live from the source data, every instructor quotation verified against the recording, and no exercise published unless its reference solution runs and passes its own tests.

  • Built from the course's own code and data, not a generic exercise set
  • Quantities computed live from the source, so a number on screen is the number in the data
  • Nothing published until its reference solution runs and passes its tests
Where it points

The roles this work is hired for.

Named because they are the roles the skills above are used in, not because we are promising you one of them.

Roles this stack is hired for
LLM EngineerAI EngineerApplied ScientistMachine Learning Engineer

These are the roles the work in this program is used in. We describe what you build and what you can do with it, and we do not promise a job, a placement, or an interview.

Simulated practice

A clock, a real task, and a room that pushes back.

The part most programs describe and never show. Here is how it actually runs, and the kind of question it runs on.

Practice runs in a timed provisioned environment, the same one the sessions run in. You get a real task and a clock, not a quiz. The session is recorded, and the review afterwards is about the reasoning you showed rather than whether you reached the expected answer.

Scenarios are written the way the question actually arrives
  • A stakeholder asks: can we trust this thing? Show me how you would answer that with a number.
  • The system gave a confidently wrong answer in a demo. Walk me through how you find out why, and what gate would have caught it.
  • We want this in front of users next month. What would you refuse to ship without?

The practice sessions are the part people underestimate and then say was the reason it worked. They are in this program from the first week.

Join the waitlist
Not into sports?

Then take the method and leave the sport.

Sports is the hook. Retrieval, evaluation, and knowing what a system can be trusted to do are the methods, and they are hired for anywhere a model touches real decisions.

  • Legal tech Retrieval over documents where a wrong citation is a real problem
  • Customer intelligence Turning unstructured conversation into something measurable
  • Enterprise search The same retrieval problem at organizational scale
  • Regulated sectors Where evaluation harnesses are the difference between shipping and not
Running alongside

The spine that does not change by phase.

One-to-one mentor sessions on your own work, fortnightly through the taught weeks and weekly through the capstone. Simulated technical interviews, recorded and reviewed. Front-office presentation practice, where a recommendation meets a room that pushes back. The Film Room once a month, alternating portfolio teardowns and interview preparation from the hiring side. See The Film Room

LLM Tools is the live track running alongside the program, so the weekly sessions and the game plan are pulling in the same direction rather than competing for your evenings.

This program is a defined credential pathway culminating in a verifiable digital badge issued through Credly, a Pearson company. We provide completion records. We are not a degree-granting or accrediting institution. If a formal credential is a requirement for you, this complements an accredited pathway rather than replacing it.

Next step

Join the waitlist.

This program is not open yet. The waitlist is the only place we announce dates, and it is one message when they land.

Join the waitlist

We will write once, when this program opens, with the dates and the application. No other list, no other mail.

Form not loading? Email [email protected] with the program name in the subject and we will add you by hand.

Every live room, the recordings, the code, and the environment, behind one door.

Join the Masterclass

See how it is taught first.

Sit in on a live session. It is the fastest way to judge any of this.