// About Novalith
Language is Structure. We Build the Systems to Read It.
A Kuala Lumpur consultancy focused on applied natural language processing for Malaysian organisations.
Back to Home// Our Story
Started from a Practical Question
Novalith was formed in 2019 by a small group of computational linguists and software engineers who had each spent years working on language technology at separate organisations across Kuala Lumpur. The question that brought them together was practical: why were so many Malaysian businesses using NLP systems that didn't understand the way Malaysians actually communicate?
Generic language models — trained almost entirely on English text from Western sources — performed poorly on Bahasa Malaysia, struggled with code-switching, and couldn't handle the regional registers that characterise professional Malaysian correspondence. The solutions on offer required significant customisation to be useful, and the expertise to do that customisation wasn't widely available locally.
Novalith was built to address that gap. We work exclusively on applied NLP for the Malaysian market, and we treat multilingual capability as a core competency rather than an edge case. Every system we deliver is designed to function in the language environment it will actually encounter.
Today our team includes linguists, machine learning engineers, and integration specialists who collaborate closely on each engagement. We are deliberate about scope — we take on projects where we believe we can do genuinely useful work, and we're straightforward when a different approach might serve a client better.
Mission
To make natural language processing useful for Malaysian organisations — by building systems that are sensitive to local language, grounded in real data, and integrated into the environments where they're actually needed.
Values
- Honesty about what NLP can and cannot do
- Rigour in methodology and evaluation
- Respect for the languages we work with
- Care with client data, always
6+
Years in NLP
50+
Projects Delivered
3
Core Services
MY
Malaysia-Focused
// Team
The People Behind the Work
Our team brings together expertise from computational linguistics, software engineering, and data science — all with direct experience in the Malaysian language environment.
Ahmad Farhan
Co-Founder & NLP Lead
Computational linguist with a background in Bahasa Malaysia morphology and multilingual corpus construction. Leads model architecture and training methodology.
Lim Wei Shan
Co-Founder & Engineering Lead
Software engineer specialising in NLP pipelines and system integration. Responsible for technical architecture and deployment across all client engagements.
Noor Raihan
Data & Linguistics Specialist
Specialises in training data curation, annotation methodology, and model evaluation. Brings deep knowledge of Malaysian vernacular registers across written and spoken forms.
Tan Kai Xiang
Machine Learning Engineer
Focuses on transformer-based model development and fine-tuning for domain-specific NLP tasks. Leads the document extraction service line.
Siti Balqis
Client Engagement Manager
Manages project scoping, delivery coordination, and ongoing client communication. Ensures technical work is grounded in each client's operational context.
Rajesh Durai
Data Engineer
Builds and maintains data pipelines for NLP input and output processing. Specialises in scalable systems that handle multilingual text at volume.
// Standards
How We Work
Our approach to every project is deliberate — methodical scoping, careful data handling, and honest evaluation at each stage.
Scoping Before Building
We don't begin development until the problem is well understood. Every project starts with a discovery phase — mapping your language data, use cases, and integration requirements.
Rigorous Evaluation
Models are evaluated on held-out data from your domain before delivery. We share evaluation results transparently — including limitations — so you can make informed decisions.
Data Privacy & Handling
Data handling agreements are established at the start of every engagement. We support on-premise deployment for sensitive data environments and operate under clear data retention policies.
Documented & Maintainable
Every system we deliver is accompanied by technical documentation. We build for maintainability — your team should be able to understand, evaluate, and extend what we build.
Collaborative Delivery
We work alongside your team rather than in isolation. Regular check-ins, shared progress views, and honest status updates are standard practice on every engagement.
Post-Delivery Support
Language environments change over time. We offer structured maintenance arrangements so models continue performing as your vocabulary, document types, or communication patterns evolve.
// Expertise
Applied NLP in the Malaysian Context
Natural language processing in Malaysia presents specific challenges that differ from those found in other markets. The coexistence of Bahasa Malaysia and English in formal and informal communication — along with Mandarin, Tamil, and regional dialects in many contexts — means that language systems need to be designed for multilingual input from the outset.
Novalith's work is grounded in domain-specific data. We've worked across financial services document processing, customer experience text analysis, public sector communication workflows, and corporate knowledge management. Each domain brings its own vocabulary, structure, and language norms — generic approaches rarely transfer well between them.
Our methodology draws on established NLP research while remaining grounded in practical delivery. We use transformer-based architectures where appropriate, traditional statistical methods where they're more interpretable and reliable for a given task, and hybrid approaches when the use case calls for it. What drives the choice is always what will work best in your environment — not a preference for any particular framework.
For organisations considering NLP adoption, we're happy to begin with a scoping conversation. We can help you understand whether a proposed use case is well-suited to NLP approaches, what data would be required, and what a reasonable timeline and scope might look like.
Interested in working with Novalith?
We're always open to a conversation about language processing challenges. If you'd like to discuss a specific project or just learn more about what NLP might do for your organisation, reach out.
Get in Touch