Kilau Neural company mission

ML capability built on what
can be measured and maintained.

We started Kilau Neural because we kept seeing the same pattern: firms deploying AI systems they could not explain, evaluate, or maintain. We work differently.

Back to Home

Kilau Neural — founded in Kuala Lumpur

Kilau Neural grew out of a straightforward observation: most AI consulting work in Malaysia at the time was either too abstract to be useful or too narrow to transfer knowledge to internal teams. Firms were paying for deliverables they could not explain to their own boards, and systems that degraded quietly once the external team left.

We set out to do the opposite. Every engagement we take is scoped to a specific business question, built with the client's team rather than alongside them, and documented in a way that leaves the firm with real capability — not a dependency on continued external support.

Our name draws on kilau — the Malay word for gleam or sheen — and the structural logic of neural networks. We like the metaphor: clarity that emerges from considered structure, not from ornamentation.

We are based at Jalan Hang Lekiu in Kuala Lumpur and work with businesses across Malaysia and the broader ASEAN region. We take a small number of engagements at any one time, because quality of work depends on attention, and attention is finite.

Scope before anything else

We refuse engagements without a defined problem. Vague mandates produce vague outputs.

Evaluation is not optional

Every system we build includes the means to measure whether it is working. We treat evaluation infrastructure as a core deliverable.

Transfer, not dependency

Our goal is for your team to be able to maintain and extend the capability we build. Knowledge transfer is structural, not cosmetic.

Honest about limits

We describe what ML systems can and cannot do. We do not compete with firms on the strength of our sales claims.

People behind the practice

NR

Nadia Ramli

Nadia leads the practice's technical engagements with particular focus on evaluation design and production ML systems. She previously held engineering roles at two Malaysian financial services firms and has worked on credit, fraud, and document processing problems across Southeast Asia.

FK

Farouk Kassim

Farouk leads senior briefings and governance engagements, drawing on fifteen years of advisory work at the intersection of technology and regulatory compliance in Malaysia and the Gulf. He has advised boards across manufacturing, logistics, and financial services on technology adoption decisions.

LW

Li Wei

Li Wei builds the data pipelines and monitoring infrastructure that sit around every model we deploy. She has designed production data systems for retail, e-commerce, and supply chain clients, and holds particular interest in the practical reliability of ML systems under real operating conditions.

How we maintain the quality of our work

PDPA Compliance

All data handling in our engagements is conducted in alignment with Malaysia's Personal Data Protection Act 2010. Data processing agreements are documented and signed before any data transfer occurs.

Evaluation-First Methodology

We do not ship model changes without a defined evaluation protocol. Our methodology requires evaluation infrastructure to be in place before any model enters production testing.

Written Scope Agreements

Every engagement begins with a written scope document agreed by both parties. Changes to scope follow a documented amendment process. Ambiguity is resolved in writing, not verbally.

Code Review & Documentation

All code produced in client engagements is reviewed internally before delivery. Documentation is written for the firm's internal technical team — not for us — and is reviewed against that standard.

Data Security Protocols

Client data is processed only in environments agreed in writing. We do not transfer client data across jurisdictions without explicit consent and appropriate contractual safeguards.

ASEAN AI Governance Awareness

We track evolving AI governance frameworks across ASEAN and incorporate relevant obligations into our briefings and build engagements. This is not legal advice; we advise clients to seek specialist counsel where specific obligations apply.

Machine learning for Malaysian operating conditions

The conditions under which ML systems operate in Malaysia are specific. Data volumes at most mid-sized Malaysian businesses are smaller than those at the large international firms whose engineering practices tend to dominate the literature. The regulatory context — particularly around personal data, financial services, and healthcare — differs from the US and European frameworks that most AI tooling is built around.

We work within these actual conditions, not around them. Our engagements are sized for the data your firm has, not for a hypothetical large corpus. Our governance recommendations reference the frameworks your firm is actually subject to.

Capability that persists after the engagement ends

The most common failure mode we encounter in post-engagement reviews is systems that degrade quietly because the internal team does not know how to maintain them. The external firm that built them has moved on. No one internally has the documentation to understand what the model does, how it was trained, or how to tell whether it is still working correctly.

We address this by treating documentation and knowledge transfer as first-class deliverables, not as optional add-ons. Our engagements close with a structured handover session and documentation written explicitly for the firm's internal team — not for a technical audience with our specific background.

We take a small number of engagements at any one time.

If you have a specific ML problem or a leadership team that needs to understand its AI choices, we would like to hear from you.

Send an Enquiry