What the Ledger Approach Gives You That a Typical Course Does Not
Longer programmes, individual review, transparent outcomes, and a structured record of your own work.
Back to HomeSix things Neurantis does differently
Individual Written Code Review
Every assignment submission receives a written review from an instructor. Not a score, not automated feedback—a set of specific observations on the code that was submitted.
Transparent Cohort Data
Completion rates and median time-on-task are published for every cohort. Prospective learners see the numbers before they decide to enrol, not after.
Negative Results Section
Neurantis maintains a record of course experiments that did not work and were withdrawn. The reasons are explained publicly. This is unusual and deliberate.
Real Compute Access
The RL course and fellowship include cluster allocation. Assignments are designed for actual hardware, not reduced for a personal laptop.
Plain Entry Requirements
Prerequisites for each programme are stated before any payment is made. There are no vague descriptions of "some experience required."
Public Research Output
Fellowship participants produce work that is released publicly, reviewed externally, and attached to their own names. The work stands as its own record.
Instructor Expertise
The instructors at Neurantis have worked in ML research and production settings, not only in education. The papers course is run by an instructor who has personally attempted to reproduce over forty published papers, including a number that proved irreproducible. That context changes what gets taught and how edge cases are explained.
Process and Method
Each programme has a defined assessment structure before enrolment begins. Learners know in advance what will be submitted, when it is due, and how it will be evaluated. There are no surprise rubric changes mid-cohort. The structure of the programme is part of what is described, not an afterthought.
Support During the Programme
Office hours are weekly. Written records of session discussions are shared within 24 hours. The RL course includes code review on every submission, not only the final project. The fellowship includes fortnightly reading group sessions with the supervising engineer throughout the 24 weeks.
Pricing Relative to Scope
The papers course at RM 520 is eight weeks with individual peer review and a final reproduction assessment. The RL course at RM 1,830 is seventeen weeks with cluster access and 15 reviewed assignments. The fellowship at RM 4,460 is twenty-four weeks with a supervising engineer, compute allocation, external critique, and publication support. The fees reflect what is provided.
Measurable Learner Outcomes
A learner who finishes the papers course has a documented reproduction record—a ledger of their own attempts on seven papers, with notes on where the reproduction succeeded and where it did not. A fellowship participant has a preregistered experiment, an ethics review record, external critique notes, and a publicly released piece of work. These are concrete, verifiable outputs.
Neurantis vs Typical Online AI Courses
| Feature | Typical Online Course | Neurantis |
|---|---|---|
| Assignment feedback | Automated scoring | Individual written review |
| Completion statistics | Not disclosed | Published per cohort |
| Compute access | Run locally or self-fund | Included for RL and fellowship |
| Failed experiments | Hidden from curriculum | Published in negative-results section |
| Entry requirements | Vague or absent | Stated plainly before enrolment |
| Ethics review | Absent or a single module | Structured review per project |
Things You Will Not Find Elsewhere
Negative-Results Archive
Neurantis maintains a public record of course experiments that were withdrawn after pilot runs, with notes on what the failure revealed. This is maintained because it is what an honest research organisation would do.
Preregistration Before Big Runs
Fellowship learners submit a written preregistration—stating their hypothesis and evaluation criteria—before any large compute allocation is approved. This is standard practice in scientific research and unusual in a learning context.
External Reviewers for Fellowship Work
Two reviewers external to Neurantis read and critique each fellowship project before public release. They are drawn from industry and academic roles. Their comments are shared with the participant in full.
Honest Timeline Estimates
Weekly hour estimates (5–7, 12–15, 18–20) are medians from completed cohorts, not marketing projections. Some learners spend more time, some less. The median is stated because it is the most useful single number.
By the Numbers
4
Years operating
214
Cohort completions
68%
Median completion rate, RL course
31
Fellowship projects released publicly
Figures cover cohorts from January 2022 through June 2025. Completion is defined as submitting all required assignments within the programme period.
See the Full Programme Details
Each programme page lists prerequisites, weekly hours, assessment structure, and what the programme does not cover.