Physical and financial systems are under growing pressure from a changing planet. Neural Earth fuses geospatial intelligence and AI to help organizations understand how environmental, social, and economic factors intersect across space and time. By transforming massive geospatial datasets into clear, actionable insight, Neural Earth enables decision-makers to anticipate risk, strengthen resilience, and plan with confidence—from individual properties to global portfolios.

We’re seeking an experienced and versatile AI/ML & Foundational Model Engineer to design, train, fine-tune, and deploy multimodal and geospatially-aware foundation models. You’ll contribute to our internal LLM stack and the intelligence layer that powers Neural Earth’s conversational AI assistant. Your work will support capabilities such as property risk reasoning, hazard detection, and spatial insight generation across imagery, text, and structured data.

This is a hands-on engineering role requiring deep experience with transformers, multimodal embeddings, and annotation or labelling pipelines. You’ll collaborate closely with our product, data, and platform teams to build domain-adapted AI systems that process text, code, imagery, and spatial datasets, and make complex risk intelligence accessible to everyone from underwriters to analysts.

• Architect and train transformer-based models, including BERT, GPT, or vision-language hybrids.
• Build workflows for supervised, unsupervised, and reinforcement learning across NLP and multi-modal tasks.
• Create high-quality datasets with robust labeling/annotation pipelines.
• Fine-tune foundation models for specific use cases (e.g., spatial data parsing, technical document summarization).
• Integrate trained models into production environments via scalable inference services.
• Monitor performance, perform evaluations, and iterate using continuous feedback loops.
• Publish internal documentation and contribute to research outputs where appropriate.
• Work with raster imagery, geospatial data, time series, video, and audio data
• Integrate databases, vector search, data lakes, and streaming data
• Build agentic AI applications for geospatial and edge computing

• 3-5+ years of hands-on experience in AI/ML engineering, with a strong portfolio of transformer or LLM-related projects.
• Proficiency with PyTorch, TensorFlow, Hugging Face, LangChain, or equivalent frameworks.
• Experience with labeling tools (e.g., Label Studio, Snorkel) and dataset versioning.
• Strong background in NLP, embeddings, tokenization, attention, and pretraining techniques.
• Understanding of model optimization techniques (e.g., quantization, distillation, LoRA).
• Ability to work with cross-functional teams on ML deployment.
• Experience with computer vision, segmentation, object recognition, and NLP

• Experience with geospatial or Earth observation data.
• Familiarity with RAG pipelines, vector databases, and multi-agent LLM orchestration.
• Contributions to open-source LLM projects or relevant academic publications.

• Up to 10%