Research Domains

We test methods to build computational thinking alongside educators. We do not just hand over software; we observe what tools survive in classrooms with intermittent electricity. We host workshops to unpack complex algorithms, helping teachers use technology as a pedagogical aid rather than a distraction. Our objective is to see educators guide their communities, confident in the digital systems they choose to adopt.

We research how to run multi-billion parameter models on standard tablets. Our engineers test quantization techniques to strip down models until they fit on edge servers. We focus on Bahasa Indonesia and regional dialects. High-speed internet is rare in the furthest reaches of the archipelago. We hypothesize that offline-first models can bridge the divide before fiber optics arrive.

We document the socio-technical impact of AI in underserved regions. Algorithmic bias looks different in a village than in a metropolis. We conduct ethnographic studies to understand how automation interacts with indigenous knowledge systems. We aim to establish a framework for rural AI that prioritizes human agency and local sovereignty over data.

We track how students interact with personalized learning paths. Our platforms use reinforcement learning to recommend content adapted to local environments. We test whether incorporating cultural metaphors improves comprehension curves for students in NTT and Sumatera. We want to ensure remote students receive careful, individualized instruction, regardless of local teacher shortages.

We deploy low-power edge nodes deep in Kalimantan. This network allows local researchers and indigenous groups to log biodiversity data without satellite links. The system manages data synchronization across nodes autonomously. This infrastructure acts as a resilient backbone for scientific field recording in heavily forested environments.

We integrate basic soil analysis and localized weather data for smallholder farmers in Java. We avoid complex dashboards. Instead, we test delivering actionable insights through simple mobile interfaces that require minimal technical literacy. We measure whether predictive warnings on water usage hold up against traditional farming intuition.

We design server nodes that function independently of grid power. We build these clusters with weather-resistant materials and standard solar arrays to survive tropical heat. We observe their uptime in community centers across Sulawesi and Kalimantan. If the edge of the network is off-grid, the knowledge base must be too.

We analyze legislative bottlenecks that slow infrastructure deployment. We speak with the Ministry of Education and Bappenas to review national roadmaps. We propose strategies like spectrum sharing to secure bandwidth for educational hubs. We map our findings to the development goals of Perpres 63/2020.