Company | Latentgrove — AI Development School

Our Story

How Latentgrove came to be

Latentgrove started in 2021 in a rented workspace in Bangsar South, where a small group of engineers and educators noticed the same pattern repeating: people were completing popular online courses and still not understanding enough to apply what they'd seen. Videos streamed fine. The gap between watching and doing was the problem.

The founders — two engineers who had both worked in applied machine learning in the region and one learning designer with a background in adult education — decided to build something narrower. Three tracks, each requiring mastery of the previous, each assessed on work produced rather than multiple-choice answers.

The name comes from a simple metaphor: a grove in which every trunk is a different learning path, and the canopy overhead is where the paths finally converge. That shared outcome — the ability to build and explain a working AI system — is what the whole structure is pointing toward.

Mission

What we are trying to do

We want people in Malaysia and the broader region to be able to build AI systems — not just describe them, not just use tools that others have built, but understand the internals well enough to make decisions about architecture, data and evaluation.

That requires more than video content. It requires individual feedback, small cohorts, real compute, and time. We organise all of those things into three connected courses and keep the rest away.

2021

Founded in KL

3

Connected tracks

<18

Per cohort

MY

Evening sessions

The People

Who you'll work with

NR

Nadia Rashid

Co-founder · Curriculum Lead

Nadia spent six years building NLP systems at organisations in KL and Singapore before returning to education. She designed the Branches and Canopy curricula and leads the fortnightly architecture reviews.

ZH

Zulkifli Hassan

Co-founder · Engineering Lead

Zul has worked on computer vision and recommendation systems and holds an MSc in Machine Learning from UTM. He maintains the infrastructure that gives learners access to GPU compute and oversees assessed project submissions.

CL

Chong Li Yen

Co-founder · Learning Design

Li Yen trained in adult education and spent years developing instructional programmes before moving into technical fields. She designed the Roots track, writes the weekly exercises, and manages the feedback process across all cohorts.

AK

Arif Kamaruzaman

Tutor · Branches & Roots

Arif runs the twice-weekly office hours for Roots and Branches and reviews submitted project code. He has a background in data engineering and writes clearly about the parts of deep learning that are most often explained badly.

SP

Siti Pahlevy

Programme Coordinator

Siti manages the operational side of each cohort — scheduling, onboarding, and the forum. She is usually the first person you hear from after an enquiry, and she keeps the details running so the tutors can focus on teaching.

RM

Reza Mohd Fauzi

Canopy Mentor

Reza is a practising ML engineer at a Kuala Lumpur fintech firm. He mentors Canopy learners through the system-build phase, providing fortnightly architecture feedback on real production considerations.

Standards

How we approach quality

These are the practices we hold across every track and every cohort — not aspirations but operating procedures.

Individual written feedback

Each weekly exercise submission receives written notes from a named tutor, not an auto-grader. Comments address the learner's actual code, not a generic rubric.

Small cohort sizes

A cohort of more than eighteen people makes individual attention impractical. We hold the number there even when demand would fill a larger group.

Data handling and privacy

Learner data — names, contact details, submission records — is held on Malaysian infrastructure and is not shared with third parties outside of legal obligations. See our Privacy Policy for detail.

Reproducible assessments

Assessed projects specify the environment and dataset so that results can be re-run by reviewers. Reproducibility is a skill we teach and a standard we hold ourselves to.

Curriculum review each cohort

After each cohort closes, the teaching team reviews what was skipped, what required extra time, and what the field has changed. Curriculum updates are applied before the next intake opens.

Honest scope of completion records

Completion records describe what was taught, how it was assessed, and the learner's standing. They do not claim to be academic qualifications. We are direct about what they are and are not.

Our Approach

What we believe about teaching AI

Most online courses in AI and machine learning have the same shape: a large library of short videos, a certificate after a quiz, and an assumption that access to information is the same thing as understanding. Latentgrove is built on a different assumption — that understanding comes from doing work and having that work reviewed by a person who knows the subject.

The three-track structure reflects how knowledge in this field actually accumulates. Python and statistical foundations are not interesting on their own, but without them, the concepts in deep learning rest on nothing. A learner who has worked through Roots arrives at Branches with the vocabulary to absorb what's there — not just the pattern-matching ability to follow a tutorial.

The Malaysian context matters to how we teach. Sessions run in the evening so that working professionals don't have to choose between their current employment and further learning. Examples and datasets are drawn from the regional context where possible. Mentors are people currently working in the industry in KL and the broader Southeast Asian region, not distant names from a different professional setting.

We are a small organisation and intend to remain one. The constraint is deliberate: it is harder to give each learner useful attention at scale, and useful attention is what distinguishes what we do from what a large platform can offer. Latentgrove's contribution to AI education in Malaysia is not breadth. It is depth, delivered carefully, one cohort at a time.

Teaching AI the way it actually works — step by step, root to canopy