Distributed Systems
Work description
Large-scale distributed systems form the backbone of modern cloud computing infrastructures and data-intensive applications. The pursuit of algorithms capable of dynamically self-adapting to variations in network conditions, workload, and other environmental factors has been a relevant line of research. Existing adaptation approaches — from dynamic parameterization at runtime to algorithm switching — are inherently limited in the space of alternatives they can explore, as correctness must be guaranteed separately for each configuration. Planned activities include: - Design and formalization of a generalized framework that decomposes distributed algorithms into independently programmable structural planes, enabling modular adaptation with correctness preservation. - Study of correctness-by-construction properties within the proposed framework, identifying invariants and structural constraints that bound the space of valid adaptations. - Development of mechanisms for optimal algorithm selection based on runtime factors such as network topology, application workload, and behavioral characteristics of the system. - Investigation of AI agent-based and optimizer-driven approaches to explore the adaptation space defined by the framework. - Evaluation of the framework's scalability and generality by mapping a diverse set of distributed algorithms onto the proposed model. - Writing a doctoral thesis in the context of the developed work. - Writing an activity report regarding the grant.
Academic Qualifications
- Enrollment in a PhD program in Computer Science or a related field.
Minimum profile required
- Knowledgeable in Distributed Systems;- Solid knowledge in consensus, replication or fault tolerance algorithms, demonstrated through academic or professional projects.
Preference factors
- Prior experience with distributed systems frameworks or middleware (e.g., BFT-SMaRt, ZooKeeper, Paxos/Raft), demonstrated through academic or professional projects; - Familiarity with adaptation models in distributed algorithms, including dynamic parameterization or algorithm switching at runtime.
Application Period
Since 21 May 2026 to 03 Jun 2026
Centre
High-Assurance Software