ppOrganisation/Company Fundació Privada I2CAT Research Field Technology » Internet technology Researcher Profile First Stage Researcher (R1) Positions PhD Positions Application Deadline 31 Aug 2026 - 00:00 (Europe/Madrid) Country Spain Type of Contract Temporary Job Status Full-time Is the job funded through the EU Research Framework Programme? Other EU programme Is the Job related to staff position within a Research Infrastructure? No /ph3Offer Description /h3pbMSCA Postdoctoral Fellowship 2026: Next-Generation of Multi-Agentic AI /b /ppThe i2CAT Foundation in Barcelona is seeking high-potential postdoctoral candidates to co-develop competitive proposals for the b2026 MSCA Postdoctoral Fellowships /b. As a premier European hub for applied research, our center focuses on generating social and economic impact in the sectors of bTelecommunications (6G), Cybersecurity, Connected Mobility, and GovTech /b. Scientific Director bDr. Xavier Costa /b and bDr. Josep Escrig /b will serve as the supervisors for this fellowship to oversee the researcher's scientific progress and professional integration. Fellows will gain access to state-of-the-art facilities, including Europe’s first 6G space laboratory, and receive dedicated proposal-writing support from our European Project Officers. /ppThe following research pillars represent the directions we consider most promising for the evolution of autonomous digital workers. These are intended as a strategic starting point; candidates are not required to address all of them and are highly encouraged to propose bvariations or entirely new research directions /b that align with their specific expertise and vision for the future of Agentic AI. /ppbResearch Pillar I: /bbMulti-Agentic AI through Continual Learning and LLM Adaptation /b /ppThis pillar investigates how LLM-powered and new generations of agents can move beyond "frozen" pre-trained weights to become truly adaptive lifelong learners. You will explore bJust-In-Time Reinforcement Learning (JitRL) /b,



a breakthrough approach that modulates a model's frozen prior toward an optimal posterior without expensive gradient updates or catastrophic forgetting . By leveraging non-parametric memory and adjusting action logits through the , agents can learn from live environmental feedback in real-time . Research will utilize state-of-the-art benchmarks like bMEAL /b (Multi-agent Environments for Adaptive Learning) to ensure these adaptive agents maintain stable cooperation even as their surroundings and roles evolve . /ppbResearch Pillar II: Quantum Multi-Agentic AI and Entanglement-Based Coordination /b /ppScaling multi-agent swarms currently faces a "communication wall" where message overhead grows quadratically with each new agent . This research line aims to shatter that wall by using bQuantum Entanglement /b as a shared substrate for coordination, effectively bypassing the need for explicit message-passing . Through bEntangled Quantum Multi-Agent Reinforcement Learning (eQMARL) /b, you will design agents whose cognitive states are physically correlated, allowing for instantaneous state alignment upon quantum measurement . This framework enables constructive interference of compatible plans while incompatible ones are canceled out, allowing massive drone swarms or 6G network slices to synchronize with zero latency . /ppbPillar III: Social Multi-Agentic AI and Cross-Stakeholder Trust /b /ppThis pillar explores the social intelligence required for Multi-Agentic AI systems where independent agents, belonging to different companies or stakeholders, must collaborate despite having different interests.



You will investigate mechanisms for bIndirect Reciprocity /b and bReputation Systems /b, allowing "digital strangers" to build trust and find human-like "win-win" collaborations in decentralized environments. By integrating mathematical bGame Theory /b and negotiation mechanisms, you will develop protocols that lead agents toward a stable bNash Equilibrium /b, ensuring fair and efficient outcomes in complex environments such as GovTech participation platforms or multi-provider 6G network slices. /ppbApplication Details and Deadlines /b /ppi2CAT provides a premier environment for researchers holding a PhD with up to 8 years of experience who meet the MSCA mobility rules. /pullibSupervisors: /b Dr. Josep Escrig and Dr. Xavier Costa** /lilibMobility Rule: /b Candidates must not have resided or carried out their main activity in Spain for more than 12 months in the 36 months (1 year of the last 3 years) immediately before the call deadline. /li /ulh3Requirements /h3pResearch Field Computer science » Other Education Level PhD or equivalent /ppSkills/Qualifications /ppMUST HAVE: /pulliPhD degree at the time of application. /liliCompliance with the MSCA mobility rule: candidates must not have resided or carried out their main activity in Spain for more than 12 months during the 36 months immediately before the call deadline. /liliRelevant scientific publications in Artificial Intelligence or related areas. /liliPrevious participation in research projects at national, European, or international level. /li /ulpSpecific Requirements /ppNICE TO HAVE: /pulliPhD in Artificial Intelligence or a closely related field. /liliStrong publication record in Artificial Intelligence. /liliPublications in Agentic AI, Multi-Agent Systems, LLM-based agents, autonomous agents, or related topics. /liliPrevious experience leading or coordinating research activities, work packages, or research projects. /li /ul /p #J-18808-Ljbffr

📌 MSCA Fellowship: Next-Generation of Multi-Agentic AI (España)
🏢 Fundacio Privada I2CAT
📍 España