What engineering teams say about working with us
Feedback from teams across Malaysia who have used Verdwell's advisory services to get a clearer view of their AI workloads.
Back to HomeFeedback from engineering teams
Wei Kang
Senior ML Engineer, Kuala Lumpur
The review session was more useful than I expected. I came in thinking we had a scheduling problem, but the session helped us see that the underlying issue was how we'd set up our job priority queues. The written summary gave us something to share with the team lead, which made the follow-up conversation a lot easier.
May 2025 — Workload Review Session
Nabilah Rashid
Engineering Lead, Johor Bahru
We did the three-week planning engagement when we were about to double our inference workload. It was genuinely helpful to have someone outside our team look at the plan before we committed to it. A few things we thought were obvious turned out to be assumptions we hadn't tested. The documented plan we got at the end became our reference for the next quarter.
April 2025 — Optimisation Planning Engagement
Suresh Velayutham
Head of AI Infrastructure, Penang
We've been on the quarterly partnership for two quarters now. What I value most is that the advisor actually remembers what we discussed last time and can track whether the changes we made had the effect we expected. It's different from a one-off engagement — there's continuity that makes the advice more useful as our workloads evolve.
May 2025 — Advisory Partnership
Lim Hui Lin
MLOps Engineer, Shah Alam
I booked the review session partly as a sanity check before presenting a resource plan to our CTO. It was worth it. A couple of the assumptions I'd made about our training job sequencing weren't as solid as I thought. The session took about two and a half hours, which was manageable, and I got the written summary three days later.
May 2025 — Workload Review Session
Azrul Hamdan
Technical Director, Johor Bahru
We used the planning engagement before starting a new batch inference system. The process was well-structured — the preparation checklist was clear about what to bring, and the discovery sessions were focused rather than open-ended. The plan we received gave us a practical sequence to follow, which helped us avoid some decisions we would have made too early.
April 2025 — Optimisation Planning Engagement
Priya Renganathan
Data Engineering Manager, Selangor
I wasn't sure what to expect from a single session, but it was more structured than I anticipated. The advisor came prepared with specific questions based on what we'd outlined in the checklist, so we got into the substance quickly. I found the observation framing in the written output useful — it was specific without being prescriptive, which gave us room to decide what to act on first.
June 2025 — Workload Review Session
How teams worked through their situations
Batch inference scaling, Johor Bahru technology company
Challenge
A team running nightly batch inference jobs was seeing unpredictable completion times as data volume grew. Some jobs would finish on schedule; others would run three or four hours longer than expected with no clear pattern. Planning the next day's operations around the output was becoming difficult.
Approach
Used the three-week planning engagement. Discovery sessions looked at how jobs were configured, what data patterns drove the variance, and how compute was allocated between stable and variable jobs. The planning phase developed a sequencing structure and a set of allocation adjustments to test.
Outcome
The team implemented the sequencing changes in the quarter following the engagement. Completion time variance dropped measurably. The team lead noted that having the plan documented meant they could check decisions against it when questions came up during implementation, rather than relying on recall from the sessions.
"What helped most was seeing the problem mapped out across the pipeline stages. Once it was laid out that way, the sequencing decisions made more sense and were easier to implement." — Engineering Lead
Training workload planning, fintech company scaling AI capability
Challenge
An engineering team preparing to move from experimental model training to regular production training cycles needed a structure for resource planning that would hold up as frequency increased. They had a rough approach in place but wanted outside input before committing to it.
Approach
Started with a single review session to assess what was in place. The session identified three areas where the current plan had untested assumptions. This led to the team commissioning a planning engagement to work through those areas specifically before launch.
Outcome
The production training cycles launched on schedule. The planning engagement had surfaced a potential bottleneck in how evaluation jobs were scheduled relative to training — addressing it in advance avoided a disruption the team would otherwise have encountered in the first production month.
"The review session felt like a low-commitment way to get an outside perspective. By the time we'd finished it, the case for the planning engagement was clear." — Technical Lead
Contact Verdwell
Phone
+60 7-228 6471Address
Jalan Wong Ah Fook 71, 80000 Johor Bahru
Hours
Mon–Fri 9:00–18:00
By the numbers
40+
Engagements delivered
4.8
Average satisfaction
3+
Years advising teams
MY
Malaysia-based
Join the teams that have brought us in for a fresh perspective
Whether it's a single session or a longer engagement, we can discuss what makes sense for where your team is right now.
Book Now