About Neurokiln
Education that states what each stage costs in time
We founded Neurokiln because most AI education conceals how demanding it is until learners are already halfway through. Every stage we run begins with a workload disclosure.
Back to HomeOur story
Why Neurokiln exists
Neurokiln started in George Town in 2022. The founding team had each spent time advising working professionals who wanted to move into AI roles — analysts, engineers and researchers who had taken online courses and found themselves underprepared for what production work required.
The recurring issue was not motivation or intelligence. It was that the programmes these learners had taken had not been straightforward about how long the difficult parts would take. People arrived at job interviews having completed courses but unable to explain their own training loops or defend their data decisions under questioning.
We built a three-stage curriculum structured like a firing schedule — where every phase has a stated temperature and duration. Learners who reach Stage 3 have written training loops from scratch, built and evaluated their own project, and worked through an actual brief from an industry partner. The completion record documents that work specifically.
Mission
What we are trying to do
Our aim is to run AI education that respects learners enough to be direct about what each stage requires. That means publishing workload figures from previous cohorts — median hours, not minimum hours — and including a dropout-reason note that explains why learners have left previous groups. We do not hide that information. We think it helps people make a better decision before they commit.
We are a small school with a single curriculum. We do not run dozens of parallel programmes. The team reviewing your milestone in Stage 2 is the same team that designed the assessment criteria. We keep the operation small enough to maintain that.
Our operating values
- State the cost before accepting payment
- Build each stage so the next one has somewhere to stand
- Give written feedback, not just scores
- Keep partner briefs real, not simulated
- Describe what the completion record is — and what it is not
The team
People behind the schedule
A small group with backgrounds in ML engineering, data science and adult education. Everyone on the team has worked in production environments before moving into teaching.
Zulaikha Ahmad
Curriculum Lead
Designed the three-stage structure and wrote the workload disclosure framework. Previously built data pipelines for a logistics firm in Penang for six years.
Rajan Nair
Stage 2 Lead Mentor
Leads the model building stage and reviews all three milestones. Spent eight years in ML engineering before joining Neurokiln in 2023.
Siti Putri
Partner Liaison, Stage 3
Manages relationships with partner organisations and coordinates the industry brief for each Stage 3 cohort. Background in project delivery and technical documentation.
Standards
How we approach education quality
Workload disclosure per stage
Before each cohort opens, we publish median and upper-quartile hours from the previous group. Figures are from learner self-reports, not our estimates.
Written exercise feedback
Every weekly exercise is returned with written notes covering what worked, what to revisit and a specific suggestion. This applies to all three stages.
Reviewed milestones in Stage 2
Three milestone submissions in Stage 2 are reviewed by a named mentor. Submissions are assessed against published criteria before the next phase begins.
Data privacy and learner records
Learner data is held and processed in accordance with Malaysia's Personal Data Protection Act 2010. Completion records are issued only to the named learner.
Verified industry partner briefs
Stage 3 briefs come from organisations that have agreed in writing to the brief format, data access scope and time commitment for the named contact who attends reviews.
Prerequisite self-check
Each stage has a self-check quiz scored locally in the browser. It is designed to help learners assess their own readiness, not to produce a pass or fail outcome that we review.
About AI education in Malaysia
Structured AI development training for working professionals in Penang and across Malaysia
Demand for AI development skills across Malaysia has grown steadily over the past four years, driven by the expanding technology sector in Penang, Kuala Lumpur and Johor Bahru. Many working professionals — particularly those in data analysis, engineering and operations roles — are looking for ways to develop practical model-building knowledge without taking extended breaks from employment.
Neurokiln addresses this by delivering all three stages entirely online, with live sessions scheduled to accommodate full-time work. The six-week preparation stage was designed specifically for people coming from spreadsheet-heavy backgrounds rather than software development, because that is the most common entry profile we see among enquiries.
The model building stage moves from first principles — writing training loops without a framework before introducing PyTorch — through convolutional and attention architectures, with a project that learners define and own. The practitioner stage then places that technical capability inside an organisational context: working with actual data constraints, actual cost limits and an actual partner who reviews the final output.
We do not describe our programme as a path to employment or a replacement for a university qualification. What we can describe precisely is what each stage covers, how long it takes at the stated pace, what feedback learners receive and what the completion record will say. Those are the things a person deciding whether to enrol can use to make a considered choice.
Take the next step
Ready to look at the stages in detail?
The solutions page has full descriptions of each stage — workload, included materials and fees — so you can assess fit before reaching out.