Context-Aware and Evolutionary AI Resilience for 6G Systems (Project GENOME)

Descrizione

Context-Aware and Evolutionary AI Resilience for 6G Systems (Project GENOME) Sintesi in italiano (traduzione automatica): L'organizzazione Telefónica Innovación Digital (TID) sta cercando un candidato per un dottorato di ricerca nel progetto GENOME, focalizzato su un framework evolutivo e consapevole del contesto per la resilienza dell'IA nei sistemi 6G. La sede del progetto è presso la Universitat Politècnica de Catalunya (UPC). Il candidato selezionato si occuperà di sviluppare meccanismi per identificare e mitigare i rischi negli ambienti di rete abilitati all'IA, utilizzando approcci Human-in-the-Loop, ottimizzazione evolutiva e tecniche di apprendimento profondo. È richiesta una laurea in ingegneria informatica o un campo correlato, oltre a una buona conoscenza della lingua inglese (C1/C2). Il lavoro prevede anche una collaborazione con NEC Laboratories Europe in Germania e la produzione di pubblicazioni scientifiche di alta qualità. Applications must be submitted exclusively through the official GENOME project website. Applications sent via LinkedIn, the Telefónica I+D website, email, or any other platform will not be considered. To ensure your candidacy is valid, you must follow the formal application process on the GENOME portal. 👉 Apply here: https://jobs.genome-msca.eu/ The GENOME Doctoral Network at Telefónica Innovación Digital (TID) is recruiting a doctoral candidate for the TID-1 research project "Context-aware and evolutionary framework for AI resilience", with PhD enrolment at the Universitat Politècnica de Catalunya (UPC). The project sits at the intersection of AI resilience, large language models, explainability, human-in-the-loop methods, and network automation for AI-native 6G and O-RAN systems. The selected researcher will develop context-aware mechanisms that identify and mitigate risks in AI-enabled network environments, with a target mean time to detect below 0.1 s. The doctoral candidate will combine Human-in-the-Loop (HITL) methods, evolutionary optimisation, prompt and context engineering, long-context processing strategies, explainable AI, and multi-objective deep reinforcement learning to design resilient AI mechanisms for operator-grade environments. The work will balance rigorous algorithmic research with practical validation on realistic telecom platforms, with a strong emphasis on scientific publications, consortium collaboration, and real-world relevance, including a planned 5-month secondment at NEC Laboratories Europe in Germany. What You Will Do (Responsibilities) Explore Human-in-the-Loop approaches that translate human language and operator intent into adaptive, security-aware system requirements and network policies. Design evolutionary and genetic optimisation methods for context embeddings and prompt adaptation in LLM-based AI resilience pipelines. Investigate large-context processing strategies, including scalable memory and decomposition methods, to handle complex security and network-management data. Develop explainable AI mechanisms that integrate with transformer-based models and help address the performance-versus-explainability trade-off. Design and validate multi-objective deep reinforcement learning architectures for identifying and mitigating risks in scenarios such as peer-to-peer and federated learning. Evaluate the framework on operator-oriented and O-RAN / vRAN experimental environments, with emphasis on securing AI-based network functions. Produce high-quality scientific publications and contribute to project deliverables D1.1, D1.2, D4.1, D4.2 and D4.3 in collaboration with GENOME partners. Language: English (C1/C2 Proficiency) Annuncio in inglese. Fonte: Euraxess (Commissione europea).