Reserved topic scholarships

Intellectual Property Notice for PhD candidates under the UniTrento-FBK Agreement 
Please read the following information carefully before submitting your application. 
Intellectual Property of Research Results.The intellectual property rights of research results generated by PhD students under scholarships within the UniTrento-FBK Agreement shall belong to FBK.
Transfer of Intellectual Property Rights. FBK will establish agreements with PhD students regarding the transfer of intellectual property rights related to their research results.
Collaboration with UniTrento. If UniTrento academic staff contribute to research results obtained through PhD scholarships funded by FBK, the determination of IP shares will be defined through separate written agreements based on each party’s contribution. PhD students are required to collaborate with UniTrento in all necessary activities related to the joint management of IP.

Department of Information Engineering and Computer Science

A1 - Trustworthy Neuro-Symbolic Integration with Multimodal Foundation Models (Project: UE HE TANGO, n.101120763; CUP E63C23000670006) (1 grant)

Multimodal foundation models are a promising venue for understanding and handling both textual and image data. However, it is well known that current foundation models can be unreliable. For instance, their outputs can be non-factual or inconsistent with respect to known logical constraints and with themselves. In turn, these problems affect potentially all down-
stream applications to which these models are applied. This project aims to address these issues by i) identifying ways in which foundation models can be made more robust by integrating ideas from neuro-symbolic AI and characterizing the fundamental limits of such integration; ii) developing novel neuro-symbolic solutions for encouraging or even ensuring foundation models comply with prior knowledge encoding, e.g., safety constraints, iii) engineering such techniques to scale to volume of data processed by foundation models, if necessary. The proposed research will explore intersections within the fields of Foundation Models, Neuro-Symbolic AI, and Explainable AI.

Contact: Andrea Passerini andrea.passerini@unitn.it

​​​​​​B1 - Algorithm/Hardware Co-design for Energy-minimal Intelligence on Resource Constrained Embedded Systems (project UE EDF RA - ARCHYTAS, n. 101167870; CUP: E63C24002180006) (1 grant)

Conventional machine learning models and algorithms are not suitable for resource-constrained edge devices due to their significant computation and resource overheads. Moreover, current edge architectures often struggle to execute these models efficiently due to control overheads within the processor datapath and the bottleneck due to the data movement between processing and memory components. This project aims to address these issues by developing novel resource-efficient inference algorithms and energy-minimal hardware architectures for embedded intelligence. This project focuses on the design of i) fast and computationally efficient inference algorithms, ii) optimization techniques that make models memory efficient while preserving accuracy, and iii) energy-efficient hardware accelerators that run algorithms by minimizing data movement within the architecture and leveraging parallelism. The proposed research will explore intersections within the fields of Tiny Machine Learning, Embedded Systems and Software, Low-Power Computing, and Computer Architecture.

Contact: Kasim Sinan Yildirim kasimsinan.yildirim@unitn.it Flavio Vella flavio.vella@unitn.it

​​​​​​B2 - Characterization and performance evaluation of interconnect architectures for HPC and AI infrastructures (Project UE HE IA - NET4EXA, n. 101175702; CUP: E63C24002170002) (1 grant)

Characterization and performance evaluation of interconnect architectures for HPC and AI infrastructures" The research activity will focus on architectures for the interconnection of computing devices for HPC (High-Performance Computing) and AI applications. These infrastructures present specific challenges, related to  the need to ensure a satisfactory quality of service (particularly guarantees of low latency) and service continuity, as well as to avoid network congestion situations. Existing architectures and those proposed in the literature will 
be studied to understand their specific aspects and to develop models capable of accurately characterizing their behavior. Based on these models and appropriate emulation and simulation environments, solutions will be studied to measure and improve performance

Contact: Fabrizio Granelli fabrizio.granelli@unitn.it Flavio Vella flavio.vella@unitn.it

​​​​​​B3 - Design and Integration of a Hardware Description Language for Optoelectronic Accelerators in Artificial Intelligence (Project UE EDF RA - ARCHYTAS, n. 101167870; CUP: E63C24002180006) (1 grant)

This research aims at the development of comprehensive software frameworks for controlling and programming photonic integrated circuits (PICs), drawing explicit parallels to how VHDL enables electronic function implementation on FPGA hardware. As photonic computing emerges as a promising solution to overcome electronic computing limitations, there remains
a significant gap in the standardization and accessibility of software interfaces for photonic hardware. The proposed research aims to develop a unified hardware description language specifically tailored for photonic systems, enabling designers to abstract complex optical phenomena into programmable components. By creating a "Photonic-HDL" equivalent to VHDL, this work will address current challenges in photonic circuit design, simulation, verification, and implementation. The research methodology will involve analyzing existing photonic design paradigms, developing suitable abstraction models, creating compiler frameworks for photonic netlists, and demonstrating practical implementations on current photonic hardware platforms.

Contact: Philippe Velha philippe.velha@unitn.it Flavio Vella flavio.vella@unitn.it

​​​​​​B4 - Exploration and optimization of heterogeneous computing architectures (Project UE EDF RA - ARCHYTAS, n. 101167870; CUP: E63C24002180006) (1 grant)

The project will investigate from a technological point of view and will develop a methodology for architecture optimization by design space exploration of heterogeneous technologies, including non-volatile devices, in-memory computing, and optoelectronics. Several efficiency metrics frequency/voltage/noise and the resulting performance metrics such as throughput, size, power and latency, will be considered, building abstractions of the components that help quickly explore the solutions and their integration together (BEOL). The exploration will be guided by constraints derived from the requirements, which feed an optimization algorithm to develop such as mixed integer linear programming. System simulation will be adopted to validate the results at an RTL level and implementation on FPGAs will allow to assess some of the metrics.

Contact: Philippe Velha philippe.velha@unitn.it Roberto Passerone roberto.passerone@unitn.it Flavio Vella flavio.vella@unitn.it

​​​​​​B5 - Low-power and batteryless wireless networking and localization (1 grant)

Low-power wireless devices are at the cornerstone of the Internet of Things (IoT), for which they provide two fundamental functionalities: the ability to interconnect devices (networking) and to determine their position in space (localization). 
The core research challenges lie in achieving high reliability and low latency with only a small energy budget. In this respect, batteryless systems push the need for low-power operation even further. These systems typically rely on supercapacitors that remove the dependance on batteries, greatly simplifying mainteinance and reducing environmental impact. However, their available energy is significantly smaller w.r.t. batteries and depleted more rapidly than it is repleted via external sources, e.g., energy harvesters. This mode of operation, alternating activity and recharging periods, results in "intermittent computing", expected to pave the way to a batteryless IoT but also significantly amplifying the challenges above. 

The PhD student will explore research themes at the intersection of networking and localization along with the challenges and limits of batteryless, intermittent computing, with the goal of contributing new techniques enabling new tradeoffs between application requirements and expected performance. The reference radio technologies for these research efforts will be ultra-wideband (UWB) and possibly Bluetooth, although others (e.g., IEEE 802.15.4, LoRa, backscatter) are also within scope.

The activities carried out can be characterized as "systems research". Novel ideas and contributions are embodied in prototypes concretely demonstrating feasibility and improvements over the state of the art. Models are used to characterize the performance of prototypes, which are then evaluated experimentally in realistic setups. In this respect, the research group offers unique assets, including a 130-node (~8000sqm) indoor UWB testbed and two accurate (mm-level) optical facilities, and along with real-world use cases from ongoing projects. 

Contact: Gian Pietro Picco gianpietro.picco@unitn.it

B6 -Multisensory creative interactions in the Musical Metaverse (Project UE HE MUSMET - Musical Metaverse made in Europe; CUP: E63C24002360006) (1 grant)

The candidate will design, develop and evaluate multi-user XR applications to support creative, multisensory, real-time interactions among musicians, among audience members and between musicians and audience members. In particular, musical haptics technologies will be integrated in XR environments, as notification methods or as strategies to enhance the musical experience. The research activities are part of the project "Musical Metaverse made in Europe" (https://musmet.eu/) funded by the European Innovation Council. 

Contact: Luca Turchet luca.turchet@unitn.it

C1 - End-to-end data-driven question-answering, reasoning and planning via customizable Reasoning Language Models (RLMs) (1 grant)

This PhD will deal with the problem of how to generate a foundation model flexible enough to be used in a variety of tasks mainly focused on reasoning, planning. The system will interact with humans (via text / prompt) and will be able to perform Question Answering on the results of its reasoning and planning. A key element of the work will be on how to avoid hallucinations and produce answers of the highest possible quality

Contact: Fausto Giunchiglia fausto.giunchiglia@unitn.it

C2 - Safe by construction Reasoning Language Models (RLMs) (1 grant)

The goal of this will be the develop techniques which will improve the performance of reasoning language Models (RLMs) in all the different phases from tokenization up to fine tuning. The overall goal will be to improve the safety of RLMs

Contact: Fausto Giunchiglia fausto.giunchiglia@unitn.it

D1 - Artificial intelligence and signal processing for planetary radar (Project: ASI JUICE RIME 2021 Bruzzone, CUP:  F65F21000950005; Project ASI EnVision fase B1 Bruzzone, CUP: F63C22000650005)  (1 grant)

The research of this grant is related to the use of signal processing and machine learning techniques for the analysis of data acquired by planetary radar sounders. Radar sounders operate from satellite platforms and acquire data related to the subsurface of celestial bodies that can results in groundbreaking science results. This activity will be developed in the framework of:
•    The Sub-surface Radar Sounder (SRS) on board the EnVision mission to Venus of the European Space Agency (ESA) (for more information refer to https://envisionvenus.eu/envision/) 
•    The Radar for Icy Moon Exploration (RIME) on board of the JUpiter ICy moons Explorer (JUICE) of the European Space Agency (see https://www.esa.int/Science_Exploration/Space_Science/Juice for more details on the mission).
The PhD research activity will include at least one of the following topics:
•    Novel methodologies and techniques based on machine learning and signal processing for the analysis radar sounder data.
•    Simulations for the analysis of the performance of the radar versus different scenarios by integrating traditional simulation techniques with machine learning approaches.
•    Test campaigns with ground penetrating radar from drone/helicopter on terrestrial sites considered as analogous of Venus and Jupiter Ice Moons subsurface.
Research will be developed at the Remote Sensing Laboratory (https://rslab.disi.unitn.it/)

Contact: Lorenzo Bruzzone lorenzo.bruzzone@unitn.it

D2 - Deep learning methodologies for the automatic analysis of satellite remote sensing images Project: ESA HRLC_CCI+ 2023 – Bruzzone; CUP E63C23002010006)  (1 grant)

The automatic analysis of image time series acquired by Earth Observation satellites is crucial for many different applications (e.g., precision farming, forestry, analysis of urban areas, monitoring of natural disasters). One of the most important applications is related to the monitoring of the land cover in relation to climate changes. In this context, even if artificial intelligence and machine learning have been widely used for the automatic analysis of remote sensing data, there are still many methodological challenges and application issues that should be addressed in the analysis of long time series of satellite data. 
The research activities of this grant are related to the development of novel methodologies based on deep learning for the automatic analysis of time series of optical (multispectral) images acquired by Earth Observation satellites. The research will be focused on the problems of the automatic classification (sematic segmentation) and change detection in multispectral images acquired by different satellite systems. The activity will consider methodological research (for the development of novel methods) and the related application to real world scenarios.
Research will be developed at the Remote Sensing Laboratory (https://rslab.disi.unitn.it/) and will be linked to the High Resolution Land Cover project developed under the leadership of RSLab in the framework of the Climate Change Initiative of the European Space Agency (https://climate.esa.int/en/projects/high-resolution-land-cover/).

Contact: Lorenzo Bruzzone lorenzo.bruzzone@unitn.it

D3 - Machine learning methodologies for the analysis of images and data acquired by planetary radars (Project: ASI Missione JUICE Bruzzone, GA 2023-6-HH.0, CUP: F83C23000070005)  (1 grant)

Space missions for the exploration of planets and celestial bodies in the Solar system are very important for the huge scientific and technological return associated with them. Some of the most challenging science objectives of these missions are related to the analysis of sub-surface processes that are of crucial importance for many different implications they have. The research activities of this grant are related to planetary radar sounders, which are instruments for the study of the subsurface of the planets of the Solar system. These radars operate from satellite platforms and acquire images related to the subsurface of celestial bodies that can results in groundbreaking science results. The activity will be focused on the development of methodologies for the processing and the automatic analysis of these data. Special emphasis will be given to methodologies that exploit the most recent developments in the framework of machine learning and signal processing.
Research will be developed at the Remote Sensing Laboratory (https://rslab.disi.unitn.it/) and will be related to the Radar for Icy Moon Exploration (RIME) on board of the JUpiter ICy moons Explorer (JUICE) of the European Space Agency (see https://www.esa.int/Science_Exploration/Space_Science/Juice  for more details on the mission).

Contact: Lorenzo Bruzzone lorenzo.bruzzone@unitn.it

Additional scholarships (published on 09.05.2025)

A10- Improving Video-Language Alignment through Synthetic Video Generation (Project:GUIDANCE (Debugging Computer Vision Models via Controlled Cross-
modal Generation) FIS2023-03251 – CUP E53C25000420001) (1 grant)
Video-language alignment (VLA) is a key multimodal task that links video content with natural language, enabling applications like video captioning and retrieval. It requires understanding both entities and their spatial-temporal relations. Current training methods using positive and negative captions generated by language models risk introducing linguistic biases. This PhD project aims to explore whether synthetic videos can help mitigate these biases and to develop methods for effectively integrating them to improve VLA performance. This research is supported by the project GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001.

Contact: Niculae Sebe niculae.sebe@unitn.it

A11- Exploring Compositionality in Vision-Language Models (Project: GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001) (1 grant)
Compositionality refers to how complex meanings are derived by combining simpler parts, a principle humans naturally use to interpret new situations. In machine intelligence, efforts have aimed to replicate this ability through tasks like sub-goals, modeling objects as part combinations, and learning compositional representations. With the rise of Vision-Language Models (VLMs), there is growing interest in exploring whether these models exhibit compositional behaviors. Previous research shows that models like CLIP represent composite concepts as linear combinations of embedding vectors. While compositionality in language has been well-explored, visual representations in VLMs remain less studied. This PhD project will investigate the compositional properties of visual embeddings in VLMs, focusing on how they represent and combine visual concepts. The research will explore how these behaviors can be used in tasks like classification, image generation, and enhancing model robustness, aiming to improve the interpretability and flexibility of VLMs by uncovering their latent compositional structures. This research is supported by the project GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001.

Contact: Niculae Sebe niculae.sebe@unitn.it

A12- Advancing Bias Detection and Mitigation Strategies for Multimodal Models (Project: GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001) (1 grant)
The widespread use of pre-trained multimodal machine learning models allows users to easily access and apply them, but many users are unaware of potential failure modes or biases that could affect performance. While model cards document such issues, not all models include them, and even when they do, they may not cover all relevant biases. Users must often assess models' weaknesses independently. Existing bias detection tools typically rely on labeled data, limiting their applicability. Recent efforts have advanced bias detection without predefined labels, but these methods still depend on task-specific data, which restricts broader automation of bias detection in models. This PhD project will focus on developing automated bias detection strategies without predefined labels and procedures for mitigating the discovered biases. This research is supported by the project GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001

Contact: Niculae Sebe niculae.sebe@unitn.it

A13 - Mitigating Bias in Text-to-Image Generation (Project: GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001) (1 grant)
Recent advancements in text-to-image generation systems have led to their widespread adoption, spurring growing interest in understanding their potential risks and embedded biases. Research has revealed significant disparities in the demographic representation of generated images, motivating the development of mitigation techniques such as textual interventions, attention-weight adjustments, and semantic guidance. This PhD project will investigate novel strategies to modify the distribution of generated content with the goal of reducing representational biases. The research will explore interventions at both the training and inference stages, aiming to develop more fair, controllable, and socially responsible generative models. This research is supported by the project GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001.

Contact: Niculae Sebe niculae.sebe@unitn.it

A14 - Foundation models for multi-modal video understanding of complex scenes (Project: GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001) (1 grant)
This PhD project will explore the development and application of foundation models for multi-modal video understanding of complex scenes. By leveraging large-scale pretraining across diverse video, audio, and textual data, we aim to build models capable of rich, contextual comprehension of video content. The focus is on creating architectures that can effectively integrate and reason over multiple modalities, enabling tasks such as complex event recognition and cross-modal retrieval. The project will investigate efficient fine-tuning methods, and benchmark evaluations on novel benchmarks to advance the capabilities and generalization of multi-modal video models. This research is supported by the project GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001

Contact: Niculae Sebe niculae.sebe@unitn.it

A15- Advancing Vision-Language Models through Commonsense Integration (Project: GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001) (1 grant)
Vision-language models have shifted AI towards more general-purpose learning, yet challenges remain in compositionality, transparency, and bridging the modality gap, particularly in contrastive learning models like CLIP. Despite significant progress, current models often struggle to holistically connect visual and textual modalities due to limited reasoning abilities. The PhD project will focus on developing self-supervised learning methods to enhance robustness, abstraction, and visual grounding of existing VLMs, integrating commonsense knowledge to improve alignment, transparency, and explainability. The research will explore different representations for reasoning and evaluate them on real-world tasks. Applications such as video question answering will serve as key benchmarks to validate the proposed methods. This research is supported by the project GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001

Contact: Niculae Sebe niculae.sebe@unitn.it

A16- Enhancing Vision-Language Models through Dynamic Adaptation and Personalization (Project: GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001) (1 grant)
Vision-Language Models (VLMs) have shown remarkable capabilities, yet they often struggle to adapt to real-world distribution shifts after deployment, limiting their effectiveness in practical applications. Overcoming this static nature is essential not only to maintain and enhance task performance over time but also to deliver a more personalized and responsive user experience. This PhD research will address a central question: How can foundational VLMs be efficiently adapted to bridge information gaps and meet evolving user needs after deployment? The project will explore innovative strategies for continual learning, model personalization, and dynamic adaptation, aiming to unlock the full potential of VLMs in real-world scenarios. This research is supported by the project GUIDANCE (Debugging Computer Vision Models via Controlled Cross-modal Generation) FIS2023-03251 – CUP E53C25000420001.

Contact: Niculae Sebe niculae.sebe@unitn.it

Fondazione Bruno Kessler (FBK)

A2 - Advancing speech recognition and understanding (1 grant)

Recent advancements in speech recognition and language technologies have significantly improved their performance across various applications: speech recognition, spoken language understanding, speaker diarization, speech analytics. However,these systems still face challenges when applied “in the wild, e.g. in typical domestic or office-like settings,  due to background noise and overlapping speech. These factors hinder the effectiveness of current speech technologies, highlighting a need for further research and development.
The project aims to investigate novel speech processing methods to address these challenges, eventually leveraging multimodal language models, to enhance performance in real operational conditions. This research will contribute to the creation of more reliable  and trustworthy speech technologies, ultimately improving user experiences and expanding the applicability of these technologies.

Contact: Alessio Brutti brutti@fbk.eu

A3 - Artificial Intelligence for Tiny, Connected Devices (1 grant)

The Internet of Things (IoT) paradigm is driving a massive increase in multimodal data generation on tiny, resource-constrained devices at the far edge of computing infrastructure. Given the challenges of limited computation, energy constraints, and communication bottlenecks, there is a growing need to process data locally while ensuring efficient and scalable artificial intelligence at the edge. TinyML has demonstrated the feasibility of embedding machine learning models on such devices. However, expanding its impact in real-world deployments requires addressing the heterogeneity of hardware, data, and resource availability in distributed scenarios.
This research will explore novel approaches for enabling AI at the edge, focusing on one or all the following aspects:
i) hardware-aware scaling, model compression, and novel approaches for diverse low-power edge devices in distributed and collaborative IoT scenarios;
ii) strategies for efficient and adaptive learning on-device or across a network of heterogeneous nodes while minimizing energy consumption and bandwidth usage;
iii) investigating how explainability and robustness can be maintained in compressed models deployed at the far edge, ensuring trustworthiness and reliability in real-world applications
This interdisciplinary research at the intersection of Artificial Intelligence, Embedded Systems, Distributed Computing, and Low-power Hardware and protocols will be tailored to the candidate's profile and interests, contributing to the development of innovative solutions for real-world challenges.

Contact: Elisabetta Farella efarella@fbk.eu

A4 - Automatic translation with large multimodal models (1 grant)

The rise of large, multimodal foundation models has driven remarkable progress in natural language processing. In text and speech translation, their growing adoption comes with quality improvements while also opening critical research directions for successful deployment and widespread access. Among the current hot research topics, three are particularly relevant to this PhD position: resource efficiency, model alignment, and model accessibility. (i) Resource efficiency aims to reduce computational demands through model compression techniques that shrink large general-purpose models, optimizing them for specific hardware (e.g., mobile devices), translation tasks, domains, or language settings. (ii) Model alignment ensures outputs are trustworthy, fair, and human-centered by integrating cultural, sociodemographic, and human factors into model design and evaluation. This includes bias-aware solutions that promote diversity and inclusivity, as well as improved evaluation methods that better capture real-world user needs. (iii) Model accessibility enhances inclusivity for individuals with visual, hearing, or cognitive impairments by integrating multimodal solutions such as sign language, speech-to-text simplification, and description-augmented subtitling, expanding the capabilities of large models. This PhD position is open to candidates with a strong interest in advancing the state of the art in any of these areas.

Contact: Matteo Negri negri@fbk.eu

A5 - Combining GeoAI and multi-modal LLM to retrieve spatial-temporal information from historical data (1 grant)

GeoAI and LLM are transforming the way we can understand and manage spatial and temporal large-scale data and territorial phenomena. The PhD project will be focused on the exploitation of historical geospatial data - from historical maps to terrestrial, aerial and satellite imagery -  alongside textual sources within a multi-modality framework. By integrating AI solutions, including LLMs, VLMs and MMLLM, the project aims to extract, structure and analyze complex geospatial information and changes over urban, forestry or glacial areas. The PhD outcomes will enhance the accessibility and understanding of historical geospatial data, boost their usage and allow for comprehensive cross-disciplinary analyses and insights into historical events and territorial transformations.

Contact: Fabio Remondino remondino@fbk.eu

A6 - Computational Social Science, Urban computing, and Computational Criminology (1 grant)

The lived experience of urban life is deeply shaped by how individuals move. In recent years, computational social scientists have increasingly leveraged mobile phone mobility data to study social and urban phenomena in a more dynamic and granular way. This data-driven approach has provided insights into accessibility to opportunities, urban resilience and evacuation patterns, socioeconomic inequality and segregation, crime, and public health dynamics.
The goal of this Ph.D. thesis is to investigate the extent to which the study of human mobility can enhance our understanding of human behavior. Specifically, this research will aim to develop computational methods that capture the complexity of urban movement and assess potential biases in the use of mobile phone data for studying mobility patterns. The ideal candidate should have a strong background in computational disciplines such as computer science, physics, data science, applied mathematics, or related areas. Additionally, an interest in interdisciplinary research at the intersection of computational social science and behavioral sciences is highly desirable. Contact person: Bruno Lepri (lepri@fbk.eu)

Contact: Bruno Lepri lepri@fbk.eu

A7 - Generative AI Agent Simulation of Behaviors, Attitudes, Interactions, and Groups (1 grant)

The general-purpose simulation of human attitudes, behaviors, and interactions could enable a laboratory for computational social scientists to test psychological, sociological, behavioral theories as well as to validate a broad set of interventions. For example, this approach could answer to how a diverse set of individuals respond to new public health policies and messages, react to product launches, or respond to major economic shocks. The goal of this Ph.D. thesis is to investigate how simulated individuals can be build leveraging large language models. The potential candidate should have a strong background in machine leaning, some experience in LLM-based social and role-playing agents and a interest on disciplines such as social sciences, and behavioral sciences. The selected candidate could be also a PhD candidate within the ELLINs doctoral school.  

Contact: Bruno Lepri lepri@fbk.eu

A8 - Human-centric machine learning and Graph Neural Network (1 grant)

Human-Centric AI aims to develop AI systems that can interact and reason in ways that align with human behaviours, ultimately augmenting and helping humans in their decision-making processes. A key challenge in this domain is building AI agents capable of understanding and integrating within complex social interactions, by learning from multimodal behavioural data and adapting to dynamic environments.
Since human interactions and behaviours naturally form structured patterns—such as social networks, communication graphs, and relational data—Graph Neural Networks (GNNs) provide a powerful framework to model these structures. This PhD position will explore the intersection of human-centric AI and graph-based learning, investigating how GNNs can enhance AI-driven simulations of human behaviour and enable adaptive, context-aware AI systems (e.g., achieving human-ai complementarity).
The ideal candidate should have a strong background in machine learning, some experience in Human-Centric AI and/or Graph Neural Networks. The selected candidate could also be a PhD candidate within the ELLIS doctoral school. Contact person: Bruno Lepri (lepri@fbk.eu)

Contact: Bruno Lepri lepri@fbk.eu

A9 - Modeling and Simulation of Urban Digital Twins (1 grant)

An Urban Digital Twin is a dynamic digital model of a city, fed by data collected from the city itself, and capable of faithfully reproducing the city behaviour through the use of advanced modelling, data science, and AI techniques. One of the key foreseen applications of Urban Digital Twins is to predict the evolution of the city, including the effects and impacts of external changes (eg, climate change) and internal processes (eg, urban transition policies and incentives). 
A challenge for the development of Urban Digital Twins is that cities are systems-of-systems, with complex interactions between physical, organisational, and social dimensions, as well as between the physical world and the digital world. Novel modeling and simulation techniques are necessary to develop Urban Digital Twins able to manage this complexity and produce reliable predictions. The candidate will be requested to work in this research area, and in particular to work on advanced agent-based modeling and simulation frameworks for Urban Digital Twins and to their validation on real scenarios concerning the adoption of Digital Twin by Italian cities. 
Besides the requirements established by the rules of the IECS school, preferential characteristics for candidates for this scholarship are:
 • master degree in Computer/Data Science, Mathematics, Physics, Electrical Engineering, Communication Engineering, or equivalents;
• knowledge in artificial intelligence, complex systems, agent-based modeling and simulation.

Contact: Marco Pistore pistore@fbk.eu 

B7 - Improving High-Speed Data Transfer with Ultra-Thin PCBs (1 grant)

The proposed topic of the thesis is related to ultra-thin PCBs, tailored for applications where intricate designs demand cutting-edge space optimization, such as in satellite payloads or large-scale scientific experiments. In detector systems, minimizing PCB thickness is often necessary to reduce dead material in the active region, where the sensor is highly sensitive to any perturbations. This is crucial for both space-based and ground-based scientific experiments.
The PhD candidate will undertake a comprehensive study encompassing (i) design and simulation, (ii) manufacturing and (iii) experimental campaigns for the high-frequency characterization (up to 15-30 GHz) of custom Printed Circuit Boards (PCBs) and various bonding schemes to chip-to-flex interconnections. 
Ultra-thin PCBs will either be manufactured in FBK via custom patent-pending techniques or by commercial standards to be used as a benchmark. The candidate will design and simulate the PCB stack, including differential pairs and controlled impedance routing.
Full process control during manufacturing will enhance the model development, allowing for the identification of specific contributions from the macroscopic geometric features (such as the shape of the metal leads) to microscopic elements like lead roughness, grain size (see Mayadas-Shatzkes model) and bonding types.
The study will explore various bonding techniques, including wire-bonding, TAB bonding and bump bonding for 3D integration. These techniques differ in materials and bonding geometries, affecting impedance and signal insertion loss. Thus, developing a computational model (e.g. using Comsol) and validating it with experimental measurements is critical for selecting the appropriate electronics design.
By validating the simulated data with VNA measurements, the investigation aims to deepen the understanding on how each factor included in the model influences the signal integrity of PCBs in high-frequency applications. Those insights will inform the design of advanced assemblies for scientific detectors in future experiments at CERN or in space missions conducted by ASI, ESA and NASA.

Contact: David Novel novel@fbk.eu Philippe Velha philippe.velha@unitn.it Luisa Petti luisa.petti@unibz.it

 C3 - Certifying model checking (1 grant)

In the field of formal verification, certifying proofs serve as compelling evidence to demonstrate the correctness of a model within a deductive system. These proofs can be automatically generated as a by-product of the verification process and are key artifacts for high-assurance systems. Their significance lies in their ability to be independently verified by proof checkers, which provides a more convenient approach than certifying the tools that generate them.
Model checking is one of the leading formal techniques used in the domain of hardware verification and synthesis, and it relies on an algorithmic exploration of the system's state space to ensure that the system behaves as expected over time. 
Modern model checking tools are sophisticated pieces of software, employing a wide range of diverse techniques to increase their performance, effectiveness, and scalability to real-world problem instances. Although often their underlying algorithms are capable of producing some form of proof, ensuring that such core proofs are compatible with the transformation and optimization
techniques applied to the input system remains a challenge.
The objective of this project is that of developing novel techniques for the reconstruction of proofs in model checking, so as to be able to certify their results. We will consider different proof formalisms and certification strategies, leveraging both interactive theorem provers and automatic decision
procedures based on propositional satisfiability (SAT) and satisfiability modulo theories (SMT). The candidate is expected to explore not only explore theoretical results, but also practical implementations on state-of-the-art model checking tools developed at FBK, and their applications to real-world problems.

Contact: Alberto Griggio griggio@fbk.eu

C4 - Failure analysis and safety assessment of complex industrial systems (1 grant)

Industrial systems are reaching an unprecedented degree of complexity. The design process has to guarantee not only the functional correctness of the implemented system, but also its dependability and resilience with respect to run-time faults. Hence, the design process must characterize the likelihood of faults, mitigate possible failures, and assess the effectiveness of the adopted mitigation measures.  
Formal methods have been increasingly used over the last decades to deal with the shortcomings of designing complex systems, in different domains. Formal methods are based on the adoption of a formal, mathematical model of the system, shared between all actors involved in the system design, and on a tool-supported methodology to aid all the steps of the design, from the definition of the architecture down to the final implementation in HW and SW. Formal methods include technologies such as model checking, an automatic technique to symbolically and exhaustively analyze all possible executions of the system in the formal model, in order to detect design flaws as early as possible. Model checking techniques have been recently extended to assess the safety and dependability characteristics of the design, and for system certification. 
The objective of this study is to advance the state-of-the-art in failure analysis and safety assessment of complex industrial systems. In particular, it will investigate extensions of existing fault propagation models and failure analysis techniques to deal with aspects such as the timing of fault propagation, the impact on system degradation, the characterization of transient and sporadic faults, and the analysis of the effectiveness of fault mitigation measures in presence of complex fault patterns. Moreover, this study will address the formal design of fault detection, isolation, and recovery (FDIR) sub-systems, techniques for diagnosis and root-cause analysis using formal methods,  anomaly detection and FDIR based on machine learning techniques. 
The developed techniques will be implemented and evaluated using tools for system-software engineering such as the COMPASS and the TASTE tools. This study is aligned with the topics investigated in various industrial projects carried out at FBK, such as the COMPASTA and AIFDIR projects, funded by the European and Italian Space Agencies.

Contact: Marco Bozzano bozzano@fbk.eu 

C5 - SMT-based software model checking (1 grant)

Techniques based on formal methods for the verification and validation of software systems are becoming increasingly important, due to the growing complexity and importance of such systems in every aspect of modern society. Despite the major progress and various examples of success stories, however, the application of formal methods, and especially automatic or fully-automatic techniques, in software remains a challenge in practice.
This project will investigate novel techniques for the application of formal methods to the design, verification, and validation of software, with particular emphasis on safety-critical application domains such as railways, automotive, avionics, and aerospace. The primary focus will be on automatic techniques combining model checking and automated theorem proving based on the Satisfiability Modulo Theories (SMT) paradigm, but other related techniques such as interactive theorem proving, abstract interpretation, or deductive verification will be considered as well. Examples of the problems tackled during the project include the formal verification of functional requirements  expressed in temporal logics, automated test-case generation, efficient handling of parametric/multi-configuration software systems and product lines, and the verification of software operating in a physical environment, subject to real-time constraints. Importantly, in addition to researching novel theoretical results, a significant part of the project activities will be devoted to the implementation of the techniques in state-of-the-art verification tools developed at FBK and their application to real-world problems in collaboration with our industrial partners.

Contact: Alberto Griggio griggio@fbk.eu

C6 - Testing methodologies for complex configurable systems (1 grant)

The escalating complexity of software systems requires the development of novel methodologies and tools for designing and testing software systems exhibiting high variability with respect to potential functional configurations in combination with different release architectures. The primary objective of this research is to investigate innovative approaches to the testing, verification, and validation of such systems, which entail the collaborative utilisation of model-based and AI-driven techniques, including formal methods, machine learning, and optimisation.

Contact: Angelo Susi susi@fbk.eu

C7 - Virtual worlds and immersive gamified experiences for positive behavior change (1 grant)

The increasing complexity of societal challenges, from climate change and energy transition to social inclusion and equity, highlights the importance of adopting innovative approaches to foster positive change and shape societies and territories of the future. In various fields — such as green transportation, biodiversity preservation, waste reduction, inclusive urban planning, and equitable access to education and resources — new strategies are essential to address these pressing issues. Virtual worlds can impact society significantly through applications such as video games, augmented reality games, and virtual reality (VR) simulations, thanks to the engagement and the perspective shift they foster. They can immerse users in unfamiliar places and situations by allowing them to take on different perspectives through avatars or virtual bodies. Research is needed to explore new ways to experience and understand our environment and society, impersonate other people or entities, and then assess how this affects cognition and behavior in the long term. The PhD candidate will research methods and techniques for creating engaging virtual worlds that promote positive behavior change through immersive, engaging experiences. In strict collaboration with a multidisciplinary research team, the candidate will develop immersive applications and contribute to design experiments that compare the approach with standard gamification techniques.
Required Candidate Skills and Competencies:
- Familiarity with at least one of the following fields: video games, gamification, serious games, VR.
- Programming experience (preferably one of C++, C#, Java, JavaScript).
- Experience with statistical analysis and data visualization software (ideally one of R Studio, Jupyter/PyCharm, Matlab, or similar).
- Good knowledge of written and spoken English.
Ideal skills (already acquired or to be acquired during the PhD):
- Experience creating virtual environments and/or 3D models (Blender, Maya, 3DSMax, or similar), for applications that run on screens or in VR.
- Experience with development tools (Unity, Unreal Engine, Godot, or similar).
- User research experience (questionnaires, interviews).

Contact: Federico Bonetti fbonetti@fbk.eu

D4 - Artificial intelligence for Earth monitoring (1 grant)

The recent Earth Observation missions like (ESA Copernicus - Sentinels, ASI PRISMA and COSMO-SkyMed, and future IRIDE constellation) make available databases of long, dense and worldwide image time series. The data have complex spatio-spectro-temporal behaviors and variability, and they show irregularities and misalignments.
Candidates will be requested to develop novel methodologies within the artificial intelligence framework for effectively and efficiently process image time series for semantic segmentation, target detection and change detection across multiannual series of data. Methodologies like foundational models, machine learning, deep learning, multitask learning, enforcement learning, etc. will be considered.
Besides the requirements established by the rules of the ICT school, preferential characteristics for candidates for this scholarship are:
•    master degree in Electrical Engineering, Communication Engineering, Computer/Data Science, Mathematics or equivalents;
•    background in artificial intelligence, image/signal processing, remote sensing, passive/active sensors.

Contact: Francesca Bovolo bovolo@fbk.eu

D5 - Bridging Space and Soil: AI-Driven Insights from Satellite and IoT Data for Smart Farming (1 grant)

The rapid advancement of Earth observation satellite constellations and the widespread deployment of IoT-based environmental sensors present an unprecedented opportunity to transform data-driven environmental modeling in precision agriculture. This PhD scholarship will explore the integration of multi-source remote sensing data with classical Machine Learning and integrative AI techniques to enhance the understanding and prediction of vegetation dynamics, soil moisture variability, and water use efficiency. By combining high-resolution satellite imagery, hyperspectral data, real-time IoT sensor networks, and data from autonomous vehicles, this study aims to develop next-generation vegetation indices and process-based models for improved agricultural monitoring and ecosystem management. The research will focus on advancing data fusion methodologies, developing scalable AI-driven models, and validating them against ground-truth datasets to enhance decision-making in precision agriculture and promote environmental sustainability.

Contact: Fabio Antonelli fantonelli@fbk.eu

D6 - Radar sounder data processing and information extraction for space exploration (1 grant)

In the context of the European Space Agency (ESA) mission JUpiter ICy moons Explorer (JUICE) to the Jovian system, we seek for candidates willing do develop methodologies for radar sounder data processing (data enhancement, semantic segmentation, denoising, content based retrieval, target detection, multitemporal analysis, radar mapping, etc.). The outcome of this activity will contribute in improving the understanding of the subsurface of planetary bodies, the correlation to history and climate as well as to a better understanding of the Earth.
The candidate will be requested to design and develop novel methodologies within artificial intelligence framework (machine learning, deep learning, quantum-based learning, etc.) for effective information extraction from radar sounder data.
Besides the requirements established by the rules of the ICT school, preferential characteristics for candidates for this scholarship are:
•    master degree in Electrical Engineering, Communication Engineering, Computer/Data Science, Mathematics or equivalents;
•    knowledge in artificial intelligence, image/signal processing, remote sensing, radar remote sensing.

Contact: Francesca Bovolo bovolo@fbk.eu