Three Programmes. Stated Prerequisites. Honest Assessment.
Each programme is described with what it covers, what it does not cover, how many hours it takes, and what it costs.
Back to HomeHow All Three Programmes Are Structured
Every Neurantis programme follows the same underlying structure: a defined set of prerequisites, a weekly hour estimate based on completed cohort data, a set of assessed exercises with written feedback, and a closing assessment. What varies is the subject matter, the depth, and the time commitment.
Assessment is based on documented work rather than on correctness alone. A reproduction that fails with a clear account of why it failed is graded as seriously as one that succeeds. This applies to the papers course directly and shapes the expectations in the RL course and fellowship as well.
All programmes include weekly contact with instructors. The format is recorded office hours (written summary distributed, no video), reading sessions, or one-on-one check-ins depending on the programme. There is no live video requirement.
Reading Research Papers as an Engineer
An eight-week course on extracting an implementation from a paper: reading the method section against the appendix, identifying what the authors omitted, reconstructing hyperparameters, reproducing a figure, and writing an honest note on where reproduction failed. Written for engineers who can train a model but find the literature opaque.
What This Programme Covers
- How to read a method section alongside its appendix and supplementary materials
- Identifying what is unstated and how to reconstruct it from context
- Six reproduction exercises on papers with public reference code
- One reproduction on a paper without any public code
- Written peer review of two classmates' reproduction notes
- Closing assessment of the learner's own full reproduction record
What This Programme Does Not Cover
- ML fundamentals or basic Python — these are prerequisites
- Paper writing or academic publishing
- Large-scale compute — exercises run on a personal machine
Entry Requirements
Ability to train a small neural network from scratch in Python. Comfort reading mathematical notation at the level of a graduate course. No prior research experience required.
Enquire About This Programme
Reinforcement Learning and Sequential Decision Making
A seventeen-week course on policy gradients, value methods, exploration, offline learning from logged data, and the evaluation problems that make this field harder than it looks. Suited to engineers with strong Python and probability who have completed a deep learning course.
What This Programme Covers
- Policy gradient methods and their variance reduction techniques
- Value-based methods: Q-learning, DQN, actor-critic architectures
- Exploration strategies and their practical implications
- Offline RL from logged data — data collection bias and evaluation
- 15 graded assignments with individual written code review
- Implementation of three algorithms from their original papers
- A learner-written simulation environment subjected to peer critique
- Cluster access for the full programme period
What This Programme Does Not Cover
- Applications involving speculation on markets or targeting of individuals
- Deep learning fundamentals — these are prerequisites
- Robotics hardware or physical simulation environments
Entry Requirements
Strong Python. Probability theory at undergraduate level. A completed deep learning course covering backpropagation and neural network training.
Enquire About This ProgrammeResearch Engineering Fellowship
A twenty-four-week fellowship for engineers who wish to work in the manner of a research group: propose a question, design an experiment, run it honestly, and publish the result whether or not it flatters. Aimed at practitioners with production experience and at postgraduate students bridging into industry.
What This Programme Covers
- Proposal development with a supervising research engineer
- Substantial cluster allocation for experiments
- Fortnightly reading group with the supervising engineer
- Experiment ledger maintained to a defined standard throughout
- Preregistration before any large compute run
- Ethics and data-provenance review of the project
- External critique from two independent reviewers
- Support in writing the work for public release
- Closing seminar with the cohort
What This Programme Does Not Cover
- Submission to journals or conferences — support ends at public release
- Any representation that work will be accepted, cited, or rewarded
- ML foundations — these must be in place before application
Entry Requirements
Production ML experience or postgraduate study. Demonstrated independent project work. Applications are reviewed individually.
Enquire About This Programme
Which Programme to Choose
Use the table below to match your current background and goals to the right entry point.
| Criterion | Papers Course RM 520 |
RL Course RM 1,830 |
Fellowship RM 4,460 |
|---|---|---|---|
| Duration | 8 weeks | 17 weeks | 24 weeks |
| Hours per week | 5–7 | 12–15 | 18–20 |
| Cluster access | Substantial | ||
| Code review | On reproduction notes | Every submission | Throughout + external |
| Public output | Personal record only | Personal record only | Released publicly |
| Best for | Engineers new to reading papers | Deep learning practitioners | Practitioners with production XP |
Standards Applied Across All Programmes
Data Privacy
Learner data is handled under Malaysia's Personal Data Protection Act. Contact and payment data is not shared with third parties for marketing purposes.
Defined Assessment Criteria
Assessment rubrics are shared with learners before any work is submitted. There are no undisclosed criteria or shifting standards mid-cohort.
Written Feedback Standard
All written feedback uses a consistent format: observation, reason, suggestion. Instructors do not use the same comment across multiple learners' work.
Ethics in Scope Definition
Applications of RL to market speculation and to individual targeting are excluded from the RL curriculum. The reasons are discussed openly in the first session of that course.
Cohort Size Limits
Cohort sizes are kept small enough that each learner receives individual instructor attention. The fellowship runs with at most twelve participants per cycle.
Programme Revision Policy
Course content changes between cohorts are documented. Learners who completed an earlier version of a programme can view a changelog describing what was updated and why.
Programme Fees
Papers Course
RM 520
Per cohort enrolment. 8 weeks. Includes all reading materials, reproduction exercises, peer review participation, and closing assessment.
EnquireRL Course
RM 1,830
Per cohort enrolment. 17 weeks. Includes cluster access, 15 assignments, code review, office hours, and simulation environment project.
EnquireFellowship
RM 4,460
Per fellowship cycle. 24 weeks. Includes supervision, substantial cluster allocation, external critique, and public release support.
EnquireAll fees are in Malaysian Ringgit (RM). Payment schedules and instalment options are discussed during the enquiry process.
Not Sure Which Programme Fits?
Send a brief note about your background and what you want to work on. We will tell you which programme is the right entry point, or whether none of them currently are.
Open an Enquiry