A steady hand on your AI pipeline
Verdwell is a Johor Bahru–based advisory practice helping engineering teams in Malaysia plan and run AI workloads with less friction and more foresight.
Back to HomeHow Verdwell came together
Verdwell grew out of a simple observation: many engineering teams that work with AI workloads are doing thoughtful, capable work, yet their compute habits are costing them more than they realise — not because of bad decisions, but because planning at the resource level often gets pushed aside when delivery pressure picks up.
The practice was established in Johor Bahru to serve teams across Malaysia who wanted structured, outside input on their AI operations — not a tool sale or a managed service, but a working advisory relationship that helps them see their own pipeline more clearly.
We focus on the operational side of AI workloads: how compute is allocated, how jobs are sequenced, how teams make decisions about resource use as workloads grow. Our work is advisory and educational — we produce written outputs that stay with the team after each engagement ends.
What we're here to do
People behind the pipeline
Ahmad Haziq
Lead Advisor
Spent several years reviewing ML infrastructure decisions at two Kuala Lumpur–based tech companies before moving into independent advisory work. Focuses on workload scheduling and compute allocation patterns.
Siew Lin
Engagement Manager
Manages the structure and delivery of client engagements, from initial scoping through to final documentation. Previously worked in technical project coordination across the Johor corridor.
Rajan Nair
Technical Analyst
Supports discovery and documentation phases, translating technical observations into structured notes and planning frameworks. Background in data engineering and operational analysis.
Standards we hold ourselves to
Defined scope for every engagement
Each service has a clear scope that we communicate before work begins. You know what's included and what the output will be — no ambiguity about what a session delivers.
Client information stays confidential
Details of your workload, architecture, and team are not shared outside the engagement. We treat access to your operational context with discretion as a baseline expectation.
Written outputs with every engagement
Advisory that exists only as a conversation doesn't survive the next sprint. Every engagement produces a written document — a summary, plan, or checklist — that your team retains.
Observation before recommendation
We take time to understand how a team's workloads actually run before suggesting anything. Recommendations grounded in observation are more relevant and easier to act on.
Scope discipline across projects
We keep our advisory focused on what was agreed. If additional areas emerge, we flag them clearly so you can decide whether to include them — rather than silently expanding the work.
Team-level view, not individual critique
Advisory is directed at how the team operates as a whole. We look at planning patterns, handoffs, and collective habits — not at evaluating individual engineers or their work.
AI workload operations, specifically
Verdwell works at the intersection of AI engineering and operational planning — a space that often gets less structured attention than either the model development side or the infrastructure configuration side. When a team is running training jobs, batch inference pipelines, or evaluation workflows at scale, the planning decisions that shape how compute is used become meaningful drivers of both cost and reliability.
Our advisory addresses questions like: how is available compute being divided across concurrent workloads, what sequencing choices are creating bottlenecks, and where are resource habits that made sense at smaller scale now creating friction? These aren't questions that require replacing infrastructure; they're questions that benefit from a considered outside view and a structured plan.
Teams we work with are typically engineering-led, working within organisations that are scaling their AI usage, and looking for practical planning support rather than a platform or a managed service. Sessions are designed to be productive for senior engineers and technical leads — people close enough to the workloads to make use of specific observations.
Being based in Johor Bahru, we work with teams across peninsular Malaysia, with sessions available both in-person and by video. The Johor corridor has seen significant growth in engineering capacity over the past several years, and we see increasing demand for this kind of operational advisory as teams move from early AI adoption to more sustained workload management.
Talk to us about your workload
If your team is spending more compute than feels right, or if you'd like a structured plan before scaling further, we're glad to have that conversation.
Get in Touch