Generative AI Engineer

Job Description: Leonardo is a global industrial group, among the main global players in Aerospace, Defence and Security that realises multi-domain technological capabilities in Helicopters, Aircraft, Aerostructures, Electronics, Cyber Security and Space. Within Leonardo's Space Division, e-GEOS , an ASI (20%) and Telespazio (80%) company, is one of the leading international operators in the Earth Observation and geospatial information sector and offers a unique portfolio of application services. We're expanding our Generative AI capabilities around next-generation Earth Observation (EO) applications, combining Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), Geospatial Foundation Models (GFMs) and agent-based workflows to transform how users access, analyse and interact with Earth Observation data, products and geo-information. Within the " Chief Technological Innovation Office " organizational unit at the Rome office we are looking for a resource to cover the position of Generative AI Engineer to help design, evaluate and implement AI-powered applications that integrate satellite imagery, geospatial datasets, technical documentation, environmental reports and operational knowledge into intelligent user experiences and decision-support systems. The role focuses on the design and implementation of production-oriented AI applications rather than foundational model research. You will work alongside software engineers, remote sensing scientists and domain experts to develop retrieval systems, AI copilots, agentic workflows and automated reasoning capabilities that leverage both structured and unstructured geospatial information. You'll be based in Rome and report to the reference technical lead, who will mentor you as you grow toward increasing ownership of our Generative AI architecture and AI-enabled products. There is a clear path toward technical ownership of our Earth Observation Generative AI stack. What you'll do Analyse, evaluate and implement state-of-the-art multi-modal and generative architectures applicable for Earth observation problems: data retrieval, preparation, model development, fine-tuning and evaluation pipelines. Design and implement Generative AI applications for Earth Observation and geospatial intelligence use cases based on cross-modal / multi-modal techniques, such as text-based object detection, visual question answering and image-text retrieval over satellite imagery. Integrate and exploit Large Language Models (LLMs) for agentic workflow execution and orchestrating multiple LLMs, Geospatial Foundation Models (GFMs), embeddings, and agents for report generation, environmental monitoring and alerting. Design and implement Knowledge Graphs of EO entities (sensors, missions, geographic features, products) for feasibility analysis and to ground LLM outputs in verified domain data. Develop Retrieval-Augmented Generation (RAG) solutions that combine Large Language Models with geospatial knowledge repositories, technical documentation, environmental datasets and operational information sources. Build agentic EO pipelines: route queries, break down tasks, call backends for data ingestion, preprocessing, chunking, metadata enrichment, indexing pipelines for structured and unstructured geospatial contents and return coherent answers. Design retrieval architectures leveraging embeddings, semantic search, hybrid search and reranking techniques. Integrate heterogeneous information sources including satellite imagery, metadata, STAC catalogues, geospatial datasets, reports, documentation, environmental observations, operational records into joint training and inference workflows. Design and orchestrate agent-based workflows combining LLMs, Geospatial Foundation Models, APIs, external tools and reasoning frameworks. Develop AI-powered assistants, copilots and automated workflows supporting environmental monitoring, geospatial analysis, report generation and operational decision-making. Experiment with fusion strategies and joint embedding spaces, benchmarking alternatives against project requirements and identifying accuracy, robustness, or cost trade-offs. Evaluate and benchmark retrieval quality, grounding accuracy, answer relevance, hallucination risk, latency and infrastructure costs. Implement prompt engineering, evaluation frameworks and testing methodologies to improve reliability and usability of AI solutions. Explore and evaluate emerging technologies, models and frameworks related to Generative AI, AI agents, reasoning systems and knowledge-augmented applications. Collaborate with platform engineers to transition prototypes into scalable, secure and production-grade services. Work closely with remote sensing scientists and domain experts to translate operational requirements into AI-enabled capabilities. Draft and review technical documentation, architectural specifications, design decisions and presentations for technical and non-technical audiences. Skills Master's degree in Engineering, Computer Science, Artificial Intelligence, Mathematics, Physics, Data Science or a related discipline; preference may be given to candidates holding a PhD. What we offer Our collective bargaining agreement is the CCNL "Industria Metalmeccanica Privata e della Installazione di Impianti" Working mode: Hybrid Contract category: Clerk Contractual type: 12-month fixed term Thirteenth month; Reward related to Business results; Welfare vouchers; Ticket of 8.50 euros Opportunities for training and continuous updating of professional skills and soft skills; Well-being: We put the economic, physical, social and psychological well-being of our people first, offering multiple solutions close to their needs. Seniority: Junior Primary Location: IT - Roma - Via Tiburtina, 965 Contract Type: Fixed term Total Base Pay Range: 35 - 44K€ Hybrid Working: Hybrid