How omnifoldaxa came to be
omnifoldaxa started in 2019 when two engineers — who had spent years building ML systems inside larger technology organisations — decided they wanted to work differently. The large-company model meant long procurement cycles, generic tooling, and projects that rarely fit the actual contours of the problem.
The name comes from origami: the idea that a flat sheet of structured data, folded correctly, can take on entirely new form and function. We apply that thinking to every engagement — taking raw, unstructured material and shaping it into something useful and maintainable.
We remain small deliberately. Our work is better when we know every client's situation in detail, when the same people who design the system are the ones building it, and when we can say no to projects that aren't a good fit.
Year omnifoldaxa was established in Kuala Lumpur
Engagements completed across Malaysia and Southeast Asia
Core AI service areas: speech, vision, and search relevance
What we're here to do
Our mission
To build AI systems that are genuinely useful inside real organisations — well-documented, honestly scoped, and handed over with the expectation that your team can run them without us.
Honest scoping
We would rather decline a project than start one that isn't likely to succeed. That means being candid about data requirements, timelines, and what AI can and can't do for a particular situation.
Practical over fashionable
We choose approaches that fit the problem, not approaches that make for impressive slide decks. Sometimes that means smaller models, simpler pipelines, or recommending against AI entirely.
Transfer of ownership
Every system we build is documented so that your team can maintain and extend it. We don't design for dependency — we design for handoff.
Data privacy by default
We treat client data carefully. We discuss data handling, residency, and processing agreements at the start of every engagement, not as an afterthought.
Who you'll be working with
Amir Hamzah
Co-founder · ML Engineering
Specialises in speech and audio processing pipelines. Previously led NLP infrastructure at a regional telecommunications group for six years.
Siow Ling
Co-founder · Computer Vision
Builds visual recognition systems for industrial and retail applications. Background in computer vision research at a Malaysian public university.
Rajan Kumar
Senior Engineer · Search & Ranking
Search relevance specialist with experience across e-commerce and enterprise knowledge management platforms across Southeast Asia.
How we maintain quality
Written scope of work
Every engagement begins with a written document specifying objectives, data requirements, deliverables, timeline, and acceptance criteria.
Offline and online evaluation
We measure model quality both against held-out test sets and, where possible, through staged rollouts with real-world metrics.
PDPA-aligned data handling
We process client data in accordance with Malaysia's Personal Data Protection Act 2010, and can sign data processing agreements on request.
Thorough documentation
Every system is documented at the architecture, API, and operational level. Your team should be able to troubleshoot and extend without calling us.
Bi-weekly progress updates
During active engagements we share written progress updates every two weeks, with clear indication of status against the agreed scope.
Handoff sessions
Each project closes with a structured handoff session where we walk your team through the system architecture, operations, and common failure modes.
AI engineering in Kuala Lumpur
omnifoldaxa operates from Bukit Bintang, one of Kuala Lumpur's most connected commercial districts, giving us direct access to clients across the Klang Valley and broader Malaysia. Our work covers three areas where AI systems deliver clear, measurable value: converting spoken language into structured data, interpreting visual information through machine learning, and improving the ranking quality of existing search infrastructure.
Malaysian organisations face a particular challenge: AI talent is scarce and expensive to retain, while off-the-shelf tools rarely handle Bahasa Malaysia, regional accents, or the specific vocabularies of local industries. omnifoldaxa closes that gap by building custom systems calibrated to your domain and your data, with documentation that keeps your team in control after we leave.
Industries we have served include call centre operations, legal documentation teams, retail e-commerce search, healthcare research, and financial services internal knowledge management. If your work produces large volumes of unstructured data — audio, images, or text — and you want to make that data more useful, we are worth a conversation.