The Applied Science Team at Giga carries out research on a range of topics, to support both products and other teams within Giga. Our skillsets cover AI and data science, more specifically: Combinatorial Optimization, Machine Learning/Deep Learning, Probabilistic Models, remote sensing, and geospatial data analysis.
Most of our work relies on several core resources, in which we are actively working on as well:
Gigaspatial: An open-source python library for geospatial processing and data harmonization.
GeoDB: An internal geospatial data base that follows the Medallion architecture
Giga_ml_utils: A python library for Machine Learning projects.
In terms of research topics, we are organized in different tracks, as follows:
AI School Mapping: we use high resolution satellite imagery to train Deep Learning models with the goal of predicting potential school geolocations (see also our AAAI-24 publication).
Procurement: we use Machine Learning and Probabilistic Models to answer questions such as how many schools are in countries where we have no school data, how many are not connected to the internet, how many do not have access to electricity, etc.
Optimization: we support infrastructure planning using Combinatorial Optimization techniques such as SAT and Constraint Programming to, for instance, calculate fiber paths to connect schools at a minimum cost.
QoS: we use Machine Learning and Time Series data analysis tools to generate insights from internet quality of service data.
Climate: we are exploring how our data can drive interventions in emergency preparedness and response under climate emergencies.
If you’re interested in working with us on research projects, reach out to jdoturodriguez@unicef.org.
