Research Focus Areas

Developing comprehensive frameworks and toolkits for educators in 3T regions to integrate AI into their pedagogy. This initiative focuses on demystifying complex AI concepts and providing practical, classroom-ready applications that work even in low-resource environments. We provide structured workshops, ongoing mentorship, and a community of practice that empowers teachers to transition from passive technology users to active AI-integrated pedagogy innovators, ensuring they can lead the digital transformation in their respective remote communities.

Researching quantization and fine-tuning techniques to run AI models on low-resource hardware in offline environments. Our core breakthrough involves optimizing multi-billion parameter models to operate effectively on standard classroom tablets and low-cost edge servers. By focusing on localized context and Bahasa Indonesia nuances, these models provide high-quality educational assistance without the need for high-speed internet connectivity, bridging the digital divide for schools in the furthest reaches of the archipelago where fiber optics are yet to arrive.

Studying the socio-technical impact of AI deployment in underserved Indonesian communities to ensure equity and safety. We investigate how algorithmic biases might affect rural populations and develop mitigation strategies that respect local cultural values and indigenous knowledge. Through deep ethnographic studies and longitudinal impact assessments, we aim to establish a framework for 'Responsible Rural AI' that prioritizes human agency, data sovereignty, and inclusive design principles, setting a global precedent for ethical technology deployment in developing contexts.

Creating personalized learning paths using AI that adapts to a student's individual pace and local curriculum needs. Our platforms utilize reinforcement learning to identify knowledge gaps and recommend remedial content tailored to the specific dialect and cultural metaphors of students in remote regions like NTT or Sumatera. This approach ensures that students in 3T areas receive the same quality of personalized instruction as their urban counterparts, despite the shortage of subject-matter expert teachers in their districts.

Implementing decentralized AI networks in the deep rainforests of Kalimantan to enable collective research databases without internet access. This project uses low-power edge nodes to create a local cloud where researchers and indigenous communities can store and analyze biodiversity data in real-time. The AI layer manages data synchronization and conflict resolution between nodes, providing a resilient infrastructure for scientific exploration in one of the world's most remote environments.

Integrating AI-driven soil analysis and crop yield prediction for smallholder farmers in underserved rural pockets of Java. By utilizing localized weather data and high-resolution satellite imagery, we provide actionable insights through simple mobile interfaces that don't require high technical literacy. The system helps farmers optimize water usage and fertilizer application, directly impacting food security and economic stability in high-density agricultural zones.

Designing and deploying independent server nodes that provide AI capabilities without grid power or connectivity. These clusters are built with weather-resistant materials and modular solar arrays, allowing them to function in extreme tropical environments. The built-in AI handles automated maintenance and power management, ensuring 24/7 availability of educational resources for community centers in the interior of Kalimantan and Sulawesi, where the electricity grid has not yet reached.

Consulting with national stakeholders to develop governance frameworks that prioritize 3T regions in digital roadmaps. Our policy research identifies legislative bottlenecks that slow down infrastructure deployment in frontier areas and proposes innovative solutions like spectrum sharing for educational AI nodes. We work closely with the Ministry of Education and Bappenas to align AI strategy with Perpres 63/2020, ensuring that the 62 priority regencies are not left behind in the generative AI era.