Building the intelligence behind better digital experiences

We are a Kuala Lumpur-based engineering team focused on AI recommendation and personalisation infrastructure for Malaysian digital businesses.

How Infrawise came to be

Infrawise grew from a straightforward observation: Malaysian online businesses were adopting recommendation technology from platforms designed primarily for North American or European user patterns. The results were inconsistent — not because the underlying algorithms were flawed, but because the data assumptions and product taxonomy structures didn't translate cleanly.

We started in 2021 as a small consultancy embedded within a Kuala Lumpur-based marketplace, working directly on recommendation layer problems. That hands-on context shaped how we think about the work. Good recommendations are not just an algorithm selection problem — they involve data architecture, catalogue structure, user behaviour modelling, and a feedback mechanism that keeps improving over time.

By 2023, we had formalised Infrawise as an independent practice, offering scoped engagements to e-commerce platforms, content publishers, and digital marketplaces across Malaysia. Our work is deliberately narrow: we do recommendation systems and personalisation infrastructure, and we do it thoroughly.

Our Mission

To make contextually appropriate AI-driven recommendations accessible for Malaysian digital businesses — with the same depth of thinking that goes into large-scale deployments, scaled to the realities of regional platform sizes and data environments.

Our Values

Honesty about what recommendation AI can and cannot do
Transparency in data handling and system design
Practicality over theoretical sophistication
Regional relevance — built for MY platforms, not adapted from abroad

The people behind Infrawise

Ahmad Firdaus

Leads recommendation algorithm development and data modelling. Previously worked on marketplace ranking systems at a regional e-commerce platform.

Soo Li Wen

Designs serving infrastructure and integration architecture. Specialises in real-time recommendation delivery and platform API integration.

Rajes Kumar

Manages ongoing analytics engagements, A/B test design, and performance optimisation. Background in conversion rate optimisation and behavioural data analysis.

How we approach quality and responsibility

Each engagement follows a consistent set of technical and ethical standards. These are not checklists — they shape how we work at every stage.

Data governance first

Data flows are documented and scoped before any modelling work begins. Access levels, retention periods, and anonymisation requirements are agreed in writing.

PDPA compliance alignment

All systems are developed with Malaysia's Personal Data Protection Act 2010 in mind. We support your legal team in understanding data processing implications.

Structured A/B testing

Recommendation quality is always tested against a baseline. We design experiments with statistical rigour and document the results clearly.

Documented, maintainable code

Every system we deliver includes technical documentation your internal team can work with. We write for handover, not dependency.

Feedback loop architecture

Systems are designed to collect implicit signals — clicks, dwell time, conversions — so models can be retrained as user behaviour evolves.

Straightforward communication

We explain tradeoffs honestly. If a particular approach isn't right for your data situation or timeline, we say so early rather than discovering it mid-project.

Recommendation intelligence for Malaysian digital commerce

The landscape for recommendation AI in Southeast Asia has developed substantially over the past several years. Malaysian e-commerce platforms, content networks, and digital marketplaces now sit on richer user interaction datasets than they did a decade ago — enough to support meaningful personalisation. The challenge is no longer data availability but rather modelling accuracy, catalogue alignment, and serving infrastructure that can keep pace with real-time browsing sessions.

Infrawise works at this intersection of data engineering and machine learning to build systems that serve contextually relevant results. Our team has hands-on experience with collaborative filtering, matrix factorisation, content embedding approaches, and hybrid architectures that blend signal types to handle sparse data environments common in growing platforms.

Operating from Kuala Lumpur, we bring a ground-level understanding of Malaysian market structures — multi-vendor marketplace dynamics, bilingual content environments, Ramadan and festive shopping patterns, and the particular way mobile-first users interact with product feeds and article recommendations. These contextual factors meaningfully affect model design and should not be abstracted away.

Our engagements are scoped to deliver usable, maintained systems — not research outputs. Every project ends with integrated infrastructure your team can operate and evolve. For platforms seeking an ongoing partner for model health and optimisation, the Analytics & Tuning service provides the continuity recommendation systems need to stay accurate as user behaviour shifts over time.

Speak with the Infrawise team

We're happy to have an initial conversation about what recommendation infrastructure could look like for your specific platform.