Summer Intern – Data Science / Earth Index… at Earth Genome · Remote

Summary: Earth Genome is looking for a technically strong data science intern to advance the core capabilities of their AI-powered platform for environmental mapping. The intern will tackle research-oriented engineering problems, including designing benchmarking frameworks and experimenting with embedding fusion strategies. Responsibilities: - Help design and build a comprehensive benchmarking suite for evaluating foundation models against Earth Index use cases, including defining metrics, curating evaluation datasets, and implementing reproducible evaluation pipelines - Experiment with approaches for combining embeddings from multiple foundation models, exploring ensemble methods, learned projections, and other fusion strategies to improve detection quality - Develop and test workflows for post-processing Earth Index detections, including clustering, filtering, confidence scoring, and spatial analysis to turn raw model outputs into clean, actionable results - Explore advanced embedding applications such as change detection (comparing embeddings across time) and multi-scale search (working across different spatial resolutions) - Write clean, well-documented, shareable code that integrates with existing Earth Index codebases - Document experimental results, design decisions, and methodology clearly enough that the team can build on your work after the internship ends Required Qualifications: - Currently pursuing (or recently completed) a graduate-level degree in computer science, machine learning, data science, or a related quantitative field - Solid Python skills and experience working with ML frameworks (PyTorch, scikit-learn, etc.) - Project experience in machine learning, particularly in areas like embeddings, computer vision, or model evaluation - Comfort working with large datasets and familiarity with concepts like vector similarity search, dimensionality reduction, or transfer learning - Experience with remote sensing or geospatial data - Methodical, experiment-driven mindset; you document what you try, not just what works - Ability to work independently, manage your own time, and communicate progress clearly in a distributed team Required Skills: Python, PyTorch, scikit-learn, Machine learning, Embeddings, Computer vision, Model evaluation, Vector similarity search, Dimensionality reduction, Transfer learning, Remote sensing, Geospatial data, Methodical mindset Benefits: Travel costs will be covered for Earth Genome offsites and any relevant workshops.