What Developers
Say About Vantyr
Feedback from developers and teams who have worked through our CUDA learning materials and programmes across Malaysia.
Back to HomeParticipant Feedback
A cross-section of responses from developers across different programmes and experience levels.
"The Parallel Basics Track filled in gaps that had been slowing me down for months. Other resources I'd tried either glossed over the memory hierarchy or assumed I already understood how warp scheduling worked. Here it's actually explained, not just mentioned. The exercises at the end of each chapter are well-designed — they require you to write something, not just read and move on."
"I went through the Applied Course as part of our team's preparation before a project involving GPU-accelerated inference. The labs were hands-on in a way that mattered — not hello-world level, but not artificially complex either. My one note is that I'd have liked more coverage of profiling tools, though the core kernel writing content was solid."
"We ran the Team Programme for six developers across two sites. The planning session before delivery was where most of the value started — they asked the right questions to understand what our team actually needed, and the sessions reflected that. The reference handbook has been used since. That's not always the case with training materials."
"What I found most useful in the Basics Track was the glossary rail. When reading about shared memory, I'd hit a term I wasn't sure about and the definition was right there contextually. It sounds like a small thing but it meant I could stay in the material instead of going off to search. Also the code is clearly tested — I didn't hit a single sample that failed to compile."
"The Applied Course covered what I needed for the project I was working on — shared memory usage, tiling patterns, avoiding bank conflicts. I came from a background in numerical C++ and the step up to GPU thinking was smoother than I expected. I'd say it's worth doing the Basics Track first if you're genuinely new to CUDA, rather than jumping straight here."
"We had a mix of experience levels in our team — some had read the CUDA docs before, others hadn't touched GPU code at all. The Team Programme handled that well. The planning call helped set expectations and the sessions were paced to bring everyone along without boring the more experienced developers. The in-person day was well-run too."
Team Programme Case Studies
Detailed accounts of how organisations approached GPU programming upskilling with Vantyr's Team Programme.
Data Infrastructure Team — Johor Bahru Fintech Company
Team Programme · April 2025 · 7 developers
The team was moving batch processing workloads to GPU-accelerated pipelines. Several developers had read about CUDA but hadn't written kernels, and there was significant variation in background across the group. A standard online course wouldn't address their specific data processing patterns.
Three sessions over two weeks — one covering the thread and memory model, one on kernel writing patterns relevant to their workloads, and one focused on understanding the output of profiling tools. All examples were adapted to data transformation patterns familiar to the team. Delivered in-person at their Johor Bahru office.
Within four weeks of the programme, three developers on the team had written and tested GPU kernels in the production pipeline. The reference handbook remained in use during that period. The team lead described the programme as giving the team "a shared vocabulary and mental model we didn't have before."
"The sessions didn't just cover CUDA in general — they covered the parts relevant to what we were building. That made the difference."
— Engineering Manager, Johor Bahru
AI Engineering Team — KL-based Product Company
Team Programme · May 2025 · 4 developers · Remote delivery
The team was building custom CUDA extensions for a deep learning framework. Two engineers had working CUDA knowledge; two were newer to GPU programming. They needed to raise the baseline of the newer developers without disrupting the workflow of the experienced ones.
Two sessions focused on the foundational gaps for the newer developers, with the senior engineers present to reinforce and add context from their experience. Remote delivery with a shared code environment for exercises. The scope was agreed to focus on memory management and kernel debugging patterns specifically.
The two newer engineers moved from reading CUDA code passively to writing and reviewing it actively within the project. The team noted a reduction in review time for GPU-related pull requests, attributed to the shared baseline established in the programme.
"Remote worked fine. The shared code environment meant we could all follow the exercises in real time, which I wasn't sure would translate."
— Senior Engineer, Kuala Lumpur
Get in Touch with Vantyr
Ready to Start?
Whether you're a developer looking to fill a specific gap or a team lead planning an upskilling effort, we'll suggest what fits your situation.
Contact Vantyr