Geospatial Data Scientist

LOGINGo back to jobsGeospatial Data ScientistBengaluru, Karnataka, IndiaApply for this jobSHARE DepartmentAI Science & ResearchJob posted onJul 16, 2026Employment typeFull-TimeApply for this jobSHARE Role: Geospatial Data ScientistEmployment Type: Full TimeEducational Qualification: B.S. or M.S. in Computer Science, Remote Sensing, Geoinformatics, Environmental Science, Physics, or a related field (or equivalent practical experience).Work Experience: •0–3 years of experience applying machine learning to geospatial, remote sensing, or other scientific/spatial data (internships and academic research projects count).Role Description: We are seeking a Geospatial AI Practitioner to join our Analytics team. This is a hands-on, build-and-execute role: you'll implement, train, evaluate, and validate machine learning and deep learning models on Pixxel's hyperspectral satellite data and allied geospatial modalities, working under the guidance of senior scientists on both production and R&D projects. You'll get exposure across the full stack of our intelligence platform, from raw data preparation and physically-informed modeling, through foundation-model fine-tuning, to the embedding- and retrieval-based systems we're building to search and reason over petabytes of imagery.Responsibilities & Duties: Build, train, and evaluate geospatial AI/ML models for applications such as crop classification, forest/biomass estimation, water quality retrieval, invasive species and land cover mapping, and change detection, under the direction of senior team members.Prepare and preprocess hyperspectral, multispectral, and SAR datasets: co-registration, atmospheric/radiometric correction, glint and shadow masking, chip generation, and label wrangling.Fine-tune and benchmark existing geospatial and foundation models (e.g., Prithvi, SatMAE, Segment Anything) on new tasks and datasets.Contribute to embedding- and segmentation-based approaches for large-scale image search and object-centric retrieval, as part of our broader platform's semantic search capabilities.Integrate multimodal data sources: hyperspectral, SAR, weather, and ground-truth/in-situ data, into modeling pipelines.Support transfer learning and domain adaptation efforts to extend models from well-labeled to label-scarce regions.Validate models rigorously against ground truth and known physical/spectral relationships, and help build out shared validation tooling and benchmarks used across the team.Write clean, well-documented, and reasonably efficient Python code, and contribute to shared libraries and pipelines used by the broader analytics team.Collaborate with data engineers, solutions scientists, and product managers to move models from notebook to production pipeline.Document methodology, maintain experiment tracking, and clearly communicate results to both technical and non-technical stakeholders.Desirable Skills & Certifications:Solid Python programming skills and working knowledge of ML libraries such as PyTorch, TensorFlow, or scikit-learn.Foundational understanding of spectral data — how absorption features, band selection, or atmospheric/surface effects influence what a sensor measures — and willingness to build deeper hyperspectral expertise on the job.Familiarity with core geospatial data handling: raster/vector formats, coordinate reference systems, and tools such as Rasterio, xarray, GDAL, or PyProj.Understanding of fundamental ML concepts (model training, validation, evaluation metrics) and willingness to learn domain-specific methods (radiative transfer, crop phenology, embedding-based retrieval) on the job.Comfortable working with large raster/imagery datasets and basic cloud or HPC compute environments.Strong communication skills and eagerness to learn from and collaborate closely with senior scientists.Coursework, research, or project experience involving hyperspectral, multispectral, or SAR satellite imagery.Exposure to cloud-based EO platforms such as Google Earth Engine,...