Based in Malaysia, Merantara provides data services for global teams developing large multimodal models, autonomous driving systems and embodied AI. We organise collection, annotation, quality assurance and delivery around local languages, everyday life and industry settings, supporting model training, local adaptation and testing.
Speech, text, image and video data reflecting Malaysia's multilingual environment. Collection covers regional accents, mixed-language communication and differences in expression across everyday life, commerce and public services.
Multilingual communication · Local expression · Multimodal understanding
Data collection and annotation for driver assistance and autonomous driving, covering local roads, road users and in-cabin interactions. Projects cover variations in road conditions, traffic behaviour and the surrounding environment to support model training and testing.
Roads and traffic · In-cabin interactions · Local scenario testing
Data from practical tasks in factories, warehouses, homes and service settings. Human demonstrations, environmental recordings and robot task data capture operating procedures, task variations and failure cases.
Human demonstrations · Robot tasks · Industry settings
Agree on the intended use, collection settings, formats and acceptance criteria. Validate the approach through a small pilot and develop clear task instructions.
Assign teams, participants, locations and tools according to task complexity and the level of standardisation. Monitor collection progress and data status.
Complete the agreed data preparation, annotation and quality assurance activities. Deliver data files and annotations with supporting quality documentation and version records.
Use client feedback from model training, testing and evaluation to guide further collection, corrections and dataset updates.
Local knowledge informs scenario selection, participant coordination and quality assurance throughout the project. Data types, permitted uses, delivery formats and acceptance procedures are agreed at the outset.
Business enquiries: [email protected]