Selected Intern's Day-to-day Resposibilities Include

• Assist with ML Model Development: Train, fine-tune, and evaluate machine learning/deep learning models on multi-modal datasets (satellite imagery, geospatial layers, audio recordings, and camera trap images) for biodiversity and carbon analytics.
• Support Data Preprocessing: Collect, clean, and preprocess raw data from various sources– satellite imagery and GIS layers, field acoustics recordings, and wildlife camera trap photos – to create model-ready datasets.
• Conduct Experimentation & Benchmarking: Conduct experiments with different model architectures and parameters; benchmark model performance on tasks (e.g. species detection accuracy, biomass estimation error) and iterate to improve results.
• Support Foundational Model Adaptation: Leverage pre-trained foundational models and frameworks (using PyTorch and HuggingFace) to accelerate development – for example, fine-tuning image classification/segmentation models and audio analysis models for our specific use cases.
• Ensure Collaboration & Reporting: Work closely with our geospatial analytics and ecology team to integrate model outputs into the platform. Document your methodologies, findings, and improvements, and contribute insights to guide decision-making in projects.
  
  

About Company: Connecting the dots - Climate. Capital. Community. Code.

At Darukaa.earth, we are a climate tech company connecting the dots between Climate, Capital, Community, and Code for democratization of nature finance and last mile impact. We leverage geospatial technology, remote sensing, blockchain, and analytics to develop high-impact environmental finance solutions.

We believe that carbon and nature-based credits are more than financial instruments they are catalysts for climate action, biodiversity protection, and community empowerment. By turning data into actionable insights, we enable businesses, policymakers, and communities to drive real environmental change.