“Using AI to Inject Noise in Java Programs”
Main adviser: Jeremy Bradbury, OntarioTech, CA
“Graph Machine Learning na Deteção Proativa de Ciberameaças integrando Visibilidade DNS e Telemetria de Host”
Coadviser: Rui Claro (Layer8)
“From Toy Examples to Real Code: A Modern, Scalable Engine for Detecting and Fixing Go Deadlocks”
Coadviser: António Ravara
“Does the Redesign Pay Off? Completing and Benchmarking a Future-Centric LSD Engine in Rust”
“Replaying What You Cannot Control: Reproducing Distributed Executions with Send Delays”
Coadviser: Eitan Farchi (IBM Haifa, IL)
“Machine Learning-Assisted Selection of Locking Strategies for Concurrent Applications”
Main adviser: Hervé Paulino
“Enhanced Stochastic Gradient Boosting for Robust Genomic Prediction in Genetic Breeding Studies”
Main adviser: Vanda Lourenço (Math Dep.)
“User Behavior Profiling via Kernel-Level Telemetry and Lightweight Edge-Inference”
Main adviser: Rui Claro, Layer8
“Extending FLeeC: A Non-Blocking Application Cache”
Coadviser: André Costa
ISO 22301 implementation thesis.
Wi-Fi aware mesh overlay P2P thesis.
Multilevel memory data storage thesis. Co-supervised (main adviser not on record).
“A Study of Data Placement and Processor Cache Prefetching in PMEM-Enabled Systems”
Coadviser: João Barreto (IST-UL)
This work aims to improve the horizontal and vertical scalability of a (JDBC) RDBMS using LSD (Lazy State Determination).
“Devault: a Decentralised Solution for Confidential Media Storage An User Centred Approach”
“Automatic Detection and Resolution of Deadlocks in GO Programs”
Main adviser: António Ravara (FCT-NOVA)
This work proposes NeoGoDDaR, a static analysis tool for the automatic detection and resolution of deadlocks in Go programs.
“Armazenamento Reativo e Persistente para Ambientes Mobile Edge Computing”
Main adviser: Hervé Paulino (FCT-NOVA)
Work done in an industrial setting.
“Controlo de Concorrência Centrado nos Dados”
Main adviser: Hervé Paulino (FCT-NOVA)
Work done in an industrial setting.
“Improving Key-Value Database Scalability with LSD: Lazy State Determination”
This work aims to improve the horizontal and vertical scalability of a KV-store using LSD (Lazy State Determination).
“Lean Data Engineering Systems: a set of Principles to Process Data Efficiently”
“OSCAR: A Noise Injection Framework for Testing Concurrent Software”
Coadviser: Jeremy Bradbury (Ontario Tech University, CA)
Design and implementation of a noise-injection tool for concurrent Java programs.
“Empowering a Relational Database with LSD: Lazy State Determination”
This work aims to improve the horizontal and vertical scalability of a (JDBC) RDBMS using LSD (Lazy State Determination).
“Telecom Churn Prediction: an Approach Towards big Data”
Coadviser: Marta Lopes (FCT-NOVA)
Telecom churn prediction thesis.
Key-value database storage thesis.
Lean Data Engineering thesis.
“OSCAR - A Java Noise Injection Framework”
“Empowering a Relational Database with LSD: Lazy State Determination”
Road traffic monitoring thesis. Co-supervised (main adviser not on record).
“Managing Population and Workload Imbalance in Structured Overlays”
“Conceção e Desenvolvimento de uma Aplicação iOS para Eliminação Assistida de Fotografias Semelhantes”
Coadviser: Fernando Birra (FCT-NOVA)
This work aims to develop an mobile application, for the iOS operating system, to help the users select and eliminate quasi-duplicated photos from the gallery of their device.
“Lazy State Determination for SQL Databases”
“Web-based Application for Assisted Elimination of Duplicate Photographs”
Coadviser: Fernando Birra (FCT-NOVA)
This work aims to develop a Web-based application to help the users select and eliminate quasi-duplicated photos from the gallery of their device.
“Runtime Tracing of Low-Code Applications @ Outsystems”
Coadviser: Herder Gregório (Outsystems)
Work done in an industrial setting.
“Energy Optimization of OpenCV Algorithms for Android”
Main adviser: Fernando Birra (FCT-NOVA)
This work aims to evaluate the energetic efficiency of a selected (small) subset of the algorithms available in the port of OpenCV for the Android operating system (in particular some that are computing intensive), and optimize some of those algorithms, maybe at the expense of some precision/quality of the final result.
“A Computing and Storage Server Infrastructure for a Mobile Application”
Coadviser: Fernando Birra (FCT-NOVA)
Grade 15/20 points. This work we plan to design and provide a server infrastructure that will be able to assist the mobile app (and the photographer) in the process of removing duplicate pho- tos, by providing services that offer features like similar image detection, image scoring and easing the process of selecting the best one that shall be kept while the others may be safely removed.
“Automated testing for provisioning systems of complex cloud products”
Coadviser: Miguel João (Outsystems)
Grade 18/20 points. This work addresses designing an efficient testing strategy that considers a microservices architecture with infrastructure provisioning capabilities while integrating it in a Continuous Integration (CI)/Continuous Deployment (CD) pipeline.
“Conceção e Desenvolvimento de uma Aplicação Android para Eliminação Assistida de Fotografias Repetidas”
Coadviser: Fernando Birra (FCT-NOVA)
Grade 17/20 points. This work proposes a software architecture to support a workflow that aims at the assisted elimination of repeated photographs.
“Suporte Aplicacional de um Workflow para Selecção Assistida de Fotografias”
Coadviser: Fernando Birra (FCT-NOVA)
Grade 18/20 points. This work aims at creating a workflow to help the user in the process of best photo selection, identifying which photos have less quality and may be safely deleted.
“Workflow Inovador de Assistência à Seleção da Melhor Fotografia”
Coadviser: Fernando Birra (FCT-NOVA)
“Best Photo Selection”
Coadviser: Fernando Birra (FCT-NOVA)
This work addresses the issue of how to help the amateur photographer to select the best photo from a set of similar photos by analysing them in technical terms. The result is a novel workflow supported by a software package, guided by user input, which will allow the sorting of the similar photos accordingly to their technical characteristics (illumination and focus) and the user requirements.
“Preventing Atomicity Violations with Contracts”
Coadviser: Carla Ferreira (FCT-NOVA)
Grade 20/20 points. This work aimed at proposing a methodolody for the development of mocular concurrent programs based in per-module contracts and a tool (gluon) that verifies that the applicaiton abides to the contracts. Co-advising by Prof. Carla Ferreira (DI-FCT-UNL). Work supported by the Synergy-VM project.
“Distributed Replicated Macro-Components”
Coadviser: Nuno Preguiça (FCT-NOVA)
“A Study on the Performance of an Object/Relational Mapping Framework”
“Lightweight Monitoring of Transactional Memory”
“A Modular Distributed Transactional Memory Framework”
Grade 20/20 points. This work aimed at developing a generic infrastructure for the executing of Java transactional memory prpgrams in a distributed setting. Work supported by the Synergy-VM project.
“A Scientific Computing Framework for Heterogeneous Environments”
“tlCell: a Software Transactional Memory for the Cell Broadband Engine Architecture”
“Static Detection of Anomalies in Transactional Memory Programs”
“TxBtrfs — A Transactional Snapshot-based File System”
Grade 18/20 points. This work aimed at detecting extending the Linux kernel module that implements the BTRFS file system to support a POSIX compliant transactional file system. Work supported by the Synergy-VM project [INForum'11].
“A Framework for Scientific Computing with GPUs”
Grade 19/20 points. This work aimed at developing a middleware for distributing scientific computations (as OpenCL kernels) over a network of nodes including CPUs and GPUs. Work supported by the RepComp project [INForum'11].
“A Transactional File System Over FUSE”
Grade 15/20.
“Practical Verification of Transactional Memory Programs (in Portuguese)”
Grade 19/20 points. This work aimed at detecting atomicity violation in transactional memory programs by static analysis of Java bytecode. Work supported by the Byzantium and Synergy-VM projects [ComSIS (5):2'09, PADTAD'11]
“tlCell: a Software Transactional Memory for the Cell Broadband Engine Architecture”
Grade 15/20 points. This work aimed at supproting the offiloading of memory transactions to the Cell processor. Work supported by the CITI research center.
“Speculative Execution by using Software Transactional Memory”
Grade 17/20. Prototype: SpecDB - Speculative Execution of Database Queries.
“SmART: An Application Reconfiguration Framework”
Coadviser: Hervé Paulino (FCT-NOVA)
Grade 18/20.
“Parallel Run-Time for CO-OPN”
Coadviser: Vasco Amaral (FCT-NOVA)
Grade 16/20.
“A Transactional File System Over FUSE (in Portuguse)”
Grade 15/20 points. This thesworkis aimed at developing a trasactional filesystem over the FUSE framework. Work supported by the CITI research center.
“Transactional File Systems”
Grade 15/20.
“SmART: An Application Reconfiguration Framework”
Grade 18/20 points. This work aimed at developing framework for the automatic configuration of applciations. Co-advising by Prof. Hervé Paulino (DI-FCT-UNL). Work supported by the VIRTU project [CSDM'10, PADTAD'10].
“Parallel Run-Time for CO-OPN”
Coadviser: Vasco Amaral (FCT-NOVA)
Grade 16/20 points. This thesis aimed at developing a run-time that allowed the automatic parallelization of CO-OPN programs.
“Transactional File Systems”
Grade 15/20 points. This work aimed at creating a library that would intercept the file system calls and provide a transactional API to the application, providing the classical database ACID properties.
His first MSc dissertation supervised, in the pre-Bologna MSc in Computer Science.
“Cooperative Memory and Database Transactions”
Grade 17/20 points. This thesis aimed at creating a unified model for memory and database transactions. Work supported by the Byzantium project [Euro-Par 2009, ComSIS 5:(2)‘08, JETC 2008].
“Consistent State Software Transactional Memory”
Grade 18/20 points. This work aimed at creating byzantine fault-tolerant applications, through a transparent layer that supports an implicit replication mechanism for RMI. [ComSIS 5:(2)‘08, JETC 2008, PADTAD'07].