8 classes · 60 min each · 1-on-1

The 8-class path to a short, real paper

Eight 60-minute, one-on-one classes — personalized to your topic and your level. Each class teaches a stage, reviews your work, and plans the next; you run the experiment between classes with my async feedback.

Not a thesis — one focused question, one clean experiment, written up like a real researcher's short paper.

$500 for the full path · Class 1 is also available standalone for $40.

01
60 min

Onboarding, topic & repo setup

$40 standalone

We pick your track — LLMs, voice, video, or RL — and set up your repo and environment so a baseline is ready to train. The lowest-risk way to start, and it's the standalone $40 Discovery Class.

Concrete example

Choose “LLM efficiency”, clone a research kit, and get a ~43M-param baseline training in ~15 min on a single GPU.

You leave withA working repo, a chosen track, and a rough question.
02
60 min

Literature review & first experiments

We read the handful of papers and PRs that actually matter for your track, pick the techniques worth testing, and run your first baseline so you have the number to beat.

Concrete example

Skim the relevant speedrun leaderboard entries and lock a 3-seed baseline val loss as your reference point.

You leave withA baseline result and a shortlist of one-line changes to try.
03
60 min

Pick one change & go deep

We commit to ONE change versus the baseline — the core of your paper — and go deep enough that you understand it and can defend it.

Concrete example

Swap ReLU² → LeakyReLU(0.5)², loop the last layers N times, or multiply the QK product by a gain.

You leave withA precise hypothesis and a working implementation of one change.
04
60 min

Experiment design & ablation

We design the minimal ablation that isolates your change — the controls, the seeds, and the single metric that decides the answer — so the result is trustworthy.

Concrete example

Sweep gain ∈ {1, 2, 5, 10}, 3 seeds each, identical tokens and config — nothing else moves.

You leave withAn experiment spec you can launch on your own.
05
60 min

Run, debug & measure

We launch the runs, hunt the silent failure modes that quietly void experiments, and confirm the numbers you collect are actually real.

Concrete example

Verify your flag is really live, re-run a seed, and build the full results table across conditions.

You leave withYour first set of clean, reproducible results.
06
60 min

Read the evidence & go deeper

We interpret signal versus noise and what the ablation proves. If it's interesting, we design one follow-up that explains WHY it works or fails. Honest null results count.

Concrete example

Add a channel-level vs head-level control to show where a measured gain actually comes from.

You leave withA defensible finding — a win, a null, or an isolated mechanism.
07
60 min

Write the paper

We write it the way researchers actually do — results first, then method, abstract, and charts — turning your experiment into a clear, short paper.

Concrete example

A 4–8 page write-up: question, method, a 3-seed results table, and an honest conclusion.

You leave withA complete first draft of your short paper.
08
60 min

Polish & publish

Final edits, a clean GitHub repo and README, and shipping it publicly. You leave able to run the next one on your own.

Concrete example

Push the repo, post the result thread, and you have real proof you can do research.

You leave withA published short paper and public repo — with your name on it.

Classes are scheduled 1-on-1 with you directly — no group cohort, no fixed weekly pace. Everything runs on a single modest GPU; all starter code and charts are provided.

Write your paper — $500, 8 classes

Founding price for the first members. Start with a single $40 class or claim a full spot.