Our Story
Machine Learning Work That Respects What Your Organisation Can Actually Do
Cendana Tech was founded in Cyberjaya in 2018 to close a specific gap: the distance between what Malaysian firms could see in their data and what they could act on. We work alongside businesses — not above them.
Back to HomeWho We Are
Grounded in Malaysian Operating Conditions
Cendana Tech began when its founders — working in an analytics role inside a Klang Valley manufacturing firm — kept encountering the same problem: consultants from abroad would propose ML approaches that assumed data conditions that simply did not exist in the business. Clean, unified data lakes. Dedicated data engineering teams. Multi-year labelling budgets. None of it matched what was on the floor.
The firm was incorporated in 2018 to offer something more grounded. We work with the data your firm already has, the staff already employed, and the problems that have a clear operational answer. We are not a research function, and we do not treat Malaysian operating conditions as a version of somewhere else with extra steps.
Our name comes from the Cendana tree — valued not for speed of growth but for density and longevity. We apply the same thinking to our work: careful, durable, and worth the time it takes.
Mission
To give Malaysian firms clear, operationally sound answers about what their data can and cannot support — and to build the systems that follow from those answers.
2018
Founded in Cyberjaya
47+
Engagements completed
3
Engagement types offered
MY
Based and operating
Our People
The Team
Small by design. Each person on an engagement has worked in the problem domain before — not just on training data sets.
Ahmad Hisham
Principal ML Engineer
Ahmad leads technical delivery across pilot engagements. He spent five years building forecasting systems inside a Penang-based electronics manufacturer before joining Cendana Tech at founding.
Nurul Rashida
Data Readiness Analyst
Nurul conducts data readiness assessments and writes the plain-language reports that come out of them. Her background is in database administration and records management for financial services firms in Kuala Lumpur.
Lim Chee Wai
Applied Researcher
Chee Wai works across classification and document processing pilots. He holds an MSc in Computer Science from Universiti Malaya and previously consulted for MSC-status companies in the Multimedia Super Corridor.
Our Standards
How We Work
The protocols that govern every engagement — not aspirational claims, but operating procedures that have been in place since the firm's founding.
Data Handling Agreement
Before any data is examined, we sign a data handling agreement specifying what can be copied, where it may reside during the engagement, and when it must be deleted.
Pre-agreed Success Criteria
We define what the model needs to do in operational terms before a line of code is written. The engagement is evaluated against those criteria, not internal benchmarks chosen after the fact.
Documentation Standard
All systems delivered include technical documentation written for the engineers who will maintain them, and a separate operational summary for the managers who will act on their outputs.
PDPA Compliance
Our processes are reviewed against the requirements of the Personal Data Protection Act 2010 (Malaysia). We advise clients on their own PDPA obligations as they relate to the data we examine.
Handover Protocol
The final week of every pilot is dedicated to handover: walkthrough sessions with your technical staff, documentation review, and written answers to questions raised during the transfer.
Plain Reporting
We do not use technical language to obscure uncertainty. If a model performs poorly in a particular area, the report names it directly and describes what would be required to improve it.
Our Position
What Machine Learning Can and Cannot Do for a Malaysian Business
The application of statistical learning systems to business problems is not new, but the marketing that surrounds it often treats these systems as more capable than they are. Cendana Tech occupies a different position: we describe what the techniques can do in terms of the data your firm actually holds, and we are specific about where a model's output should be checked by a person before acting on it.
For a Malaysian operating business — whether in manufacturing, financial services, retail, or a professional practice — the relevant questions are usually about classification and forecasting. Can this system tell me which invoices are likely to be late? Which products are most likely to be returned? Which applicants fall into a risk category worth examining further? These are answerable questions, given data of reasonable quality and sufficient volume.
What these systems cannot do is replace the judgement required for novel situations, edge cases, or decisions that carry regulatory weight. A model trained on historical data will perform poorly when the underlying conditions change. It will produce outputs with confidence when it has encountered inputs far from its training distribution. Our assessments name these limits as plainly as they name the capabilities.
Cendana Tech has conducted engagements across the Klang Valley, Penang, and Johor Bahru. In each case, the value came not from the sophistication of the technique but from the care taken at the front end: understanding the data, understanding the decision being supported, and scoping the work accordingly. That is the practice we have built, and it is what we bring to each new engagement.
Ready to Discuss Your Situation?
We are available for an initial conversation at no cost. Describe your data situation and the problem you are trying to address, and we will give you an honest assessment of whether and how we can help.
Contact Cendana Tech