◈ About Us
The People Behind
Novathread
We are a small, focused team of AI practitioners working out of Sai Ying Pun. Our work is practical by design: we solve specific organizational problems with AI, and we do not oversell what the technology can do.
← Back to Home◈ Our Story
Built in Hong Kong, for Organizations Here
Novathread began when a small group of engineers and analysts, working across financial services and research technology in Hong Kong, found themselves repeatedly solving the same class of problem: organizations had data that contained useful signals, but no reliable way to surface them systematically. The tools available were either too generic, too expensive, or too dependent on infrastructure no mid-sized organization could realistically maintain.
The name reflects how we think about this work — threads of data woven together through deliberate structure, much like a circuit connecting functional nodes across a board. We are not trying to build the largest AI platform. We are trying to build the right system for the organization in front of us.
Since establishing our Sai Ying Pun office, we have worked with compliance teams, legal departments, and technology leads across Hong Kong and the wider Asia-Pacific region. Our client relationships tend to be long ones, built on transparency about what AI can and cannot do, and on a genuine interest in making the handover successful.
Mission
To give organizations in Hong Kong access to capable, well-integrated AI systems — without the noise, the vendor lock-in, or the inflated promises that often surround the category.
Values
- Clarity over complexity
- Impartiality in advice
- Sustainable, maintainable builds
- Respect for data privacy obligations
Who We Work With
Financial institutions, legal practices, insurance providers, research organizations, and regulated platforms — organizations where data accuracy, auditability, and privacy matter from day one.
◈ The Team
People Who Build This
A deliberately compact team, each person carrying depth in their discipline and a shared disinterest in AI hype.
Li Kam-Fung
Co-Founder & Technical Lead
Twelve years building data systems across financial services. Li leads system architecture and oversees the technical quality of every engagement.
Sophia Ramirez
Co-Founder & Client Advisory
Former technology strategist with a background in procurement and vendor evaluation. Sophia leads the advisory practice and manages all client relationships.
Marcus Chan
ML Engineer
Specialist in NLP and anomaly detection. Marcus handles model development, evaluation frameworks, and the harder engineering problems in each engagement.
◈ Standards
How We Operate
These are not aspirational values — they are working practices we apply on every engagement.
Data Privacy by Design
All systems are architected around compliance with Hong Kong's PDPO from the outset. We do not treat privacy as an afterthought or an add-on.
Transparent Model Evaluation
We share evaluation metrics, error analysis, and confidence ranges with clients before deployment. You should understand how the model performs before relying on it.
Clean, Documented Handovers
Every engagement concludes with structured documentation, system diagrams, and a handover period to ensure your team can operate and maintain what we've built.
Impartial Advisory
We have no vendor partnerships, reseller agreements, or commercial arrangements that could bias our recommendations. Our guidance reflects your interests, not ours.
Iterative Delivery
We work in defined phases with review checkpoints, giving you the opportunity to redirect before significant work has been invested in the wrong direction.
Direct Access to Practitioners
You communicate directly with the engineers and advisors working on your project. There is no account management layer between your questions and the people with the answers.
◈ Our Approach
AI Work That Fits Organizational Reality
Organizations working with sensitive data — in financial services, legal research, or regulated compliance functions — face a particular challenge when considering AI adoption. The potential is real, but so are the risks of deploying systems that underperform, produce unexplainable outputs, or create new compliance exposure. Novathread operates specifically in this space.
Our fraud detection capability draws on pattern analysis across transaction data, device signals, and behavioral context. Unlike rule-based systems that require constant manual tuning, our models update continuously as the threat landscape shifts — while remaining auditable for your compliance team. We integrate with your existing processing infrastructure rather than replacing it.
Document analysis at scale, particularly for semantic similarity, sits at the intersection of language understanding and domain knowledge. Our systems are trained and validated on the document types your organization actually works with — legal filings, patent applications, regulatory submissions, or research outputs — rather than on generic text. The confidence scores and highlighted passages returned by our tools are designed to support, not replace, expert review.
For organizations still forming their AI strategy, the procurement guidance service provides an independent, technically grounded perspective. We help convert business requirements into specific technical specifications, create evaluation frameworks, and assess vendor proposals. This service is frequently engaged by legal, procurement, or technology leadership teams who want to make informed decisions without relying solely on vendor demonstrations.
◈ Work Together
A Considered Approach Starts With a Conversation
We take time to understand what you are trying to achieve before proposing anything. An initial call or meeting carries no obligation.
Contact Novathread