Building Clearer Paths
into GPU Programming
Vantyr was established in Johor Bahru with a clear focus: give developers in Malaysia a well-structured way to learn CUDA parallel programming without the usual gaps and guesswork.
Back to HomeWho We Are
Vantyr started when a small group of engineers in Johor Bahru noticed that most publicly available CUDA learning material skipped the parts that actually trip people up. Memory hierarchy, thread divergence, warp scheduling — topics documented somewhere, but rarely presented in a way that builds coherent understanding.
Rather than produce another video series of loosely connected demos, we chose to work from first principles: what does a developer need to understand before writing a kernel that performs well? What mental model has to be in place before optimisation decisions make sense? That thinking shaped how we structured our tracks.
Our materials read more like a technical textbook than a tutorial collection. Code panels sit beside explanations. Glossary terms are defined before they're used. Exercises are tied directly to the concept just introduced, not added as an afterthought.
Operating out of Johor Bahru also means we're well-positioned to work directly with engineering teams based in southern Malaysia — in-person sessions, on-site programme delivery, and follow-up conversations that don't require a flight.
struct Company {
string name = "Vantyr";
string location = "Johor Bahru, MY";
string focus = "CUDA Education";
bool in_person = true;
bool remote = true;
};
Our Mission
To make GPU parallel programming accessible to working developers through materials that explain the fundamentals clearly — without glossing over the parts that take the most time to learn on your own.
What We Value
- Accuracy over simplification
- Progression that doesn't skip steps
- Code that actually compiles and runs
- Plain language before jargon
The People Behind Vantyr
A compact team with backgrounds in GPU computing, software engineering, and technical instruction.
Razif Amirul
Lead Curriculum Developer
Spent several years in GPU compute roles before shifting focus to education. Designs the core content structure and code examples across all Vantyr tracks.
Nurul Tahira
Applied Labs Instructor
Leads the hands-on lab sessions in the Applied Course and Team Programme. Builds exercise sets based on real-world GPU workloads from her background in AI infrastructure.
Wei Kang
Technical Review & Operations
Reviews all code examples for correctness and reviews materials for clarity before publication. Also handles programme logistics and client coordination for team engagements.
How We Maintain Quality
Standards that apply to every track, course, and programme we produce.
All Code Is Tested
Every code example in our materials compiles and runs against current CUDA toolkit versions. We don't include illustrative snippets that wouldn't actually work on real hardware.
Technical Peer Review
New content goes through an internal review pass before release. The goal is catching both factual errors and places where explanation could be clearer or more precise.
Regular Content Updates
CUDA evolves. We revisit materials periodically when new toolkit versions or architecture features introduce meaningful changes to how things work or should be written.
Data Privacy Standards
Participant data from courses and team programmes is handled with care. We collect only what's needed for delivery and don't share it with third parties outside of stated service providers.
Participant Feedback Loops
Feedback from course and programme participants feeds back into how we revise exercises and explanations. Sections that consistently confuse people get rewritten, not left as-is.
Plain Language First
We write for developers, not for reviewers. Technical accuracy and clear prose aren't in tension — we aim for both, and we favour straightforward sentences over language that sounds impressive but reads slowly.
CUDA Programming Education for Malaysian Developers
GPU parallel programming is increasingly relevant to developers working on machine learning pipelines, simulation workloads, and data-intensive applications. CUDA remains the primary programming model for NVIDIA hardware, and developers who can write and reason about GPU kernels have a broader set of options when working on performance-sensitive systems.
Vantyr provides learning materials that cover CUDA from foundational to applied levels — thread and block structure, shared memory usage, memory bandwidth considerations, warp-level execution, and kernel optimisation approaches. The materials are written for developers with C or C++ backgrounds who want to understand what's happening inside the GPU, not just copy-paste patterns from samples.
Our team programme adapts this content for organisations. Rather than expecting engineers to self-study, the engagement includes planning sessions to understand the team's current baseline, structured delivery across multiple sessions, and a reference handbook the team keeps. Companies with developers working on AI training infrastructure, rendering pipelines, or scientific computation will find this relevant.
Located in Johor Bahru, Vantyr works with clients across Johor and beyond — in person where practical, remotely where preferred. The combination of structured written materials and direct engagement distinguishes what we offer from generic online courses or documentation-heavy reference material.
Want to Know More?
If you'd like to understand which track or programme suits your situation, we're happy to have that conversation. No hard sell — just a straightforward discussion.
Contact Vantyr