About

Dilma Da Silva is a Professor in the Department of Computer Science & Engineering at Texas A&M University. She holds a Ph.D. in Computer Science from Georgia Institute of Technology, obtained in 1997, and both her M.S. and B.S. degrees in Computer Science from Universidade de São Paulo, completed in 1990 and 1986 respectively. Her research interests include cloud computing, operating systems, distributed computing, and high-end computing. She has received numerous awards and honors, including being named an ACM Distinguished Speaker from 2009 to 2014, an ACM Distinguished Scientist in 2011, and receiving the IBM Research Division Award in 2011. Her work has been recognized for its research productivity, and she has contributed significantly to her field through various publications and research projects.

Research topics

• Computer Science
• Sociology
• Artificial Intelligence
• Programming language
• Information Retrieval
• Political Science
• World Wide Web
• Mathematics
• Pedagogy
• Law
• Software engineering
• Psychology
• Library science
• Database
• Mathematics education
• Multimedia
• Computer architecture
• Data science
• Operating system

Selected publications

• Teaching Algorithms to Indigenous Students of Brazil's Amazon

  2025-02-18

  articleOpen accessSenior author

  The Constitution of Brazil and its subsequent laws have established various rights and protections for Indigenous peoples, among them the right to Indigenous schools where their culture and native language must be taught, learned, and preserved as something alive and essential to their well-being. Brazil's National Digital Education Policy, which mandates the teaching of computing in K-12 education, is a recent development not yet implemented in indigenous schools, where access to computers and the Internet is still quite limited. To promote the inclusion of indigenous populations in higher education, the University of Brasília (UnB), in collaboration with Brazil's National Foundation for Indigenous Peoples, has created an admission pathway dedicated to students from these populations. In 2022, UNB's Computer Science Department welcomed its first three Indigenous students from the Ticuna community in the Amazon region of Brazil. The Ticuna people represent the largest indigenous ethnic group in Brazil. Ticuna students, computer science professors, and computer science students at UnB have collaborated to address the gap in the K-12 teaching of computing in Ticuna communities. This work describes the materials created by the indigenous students for teaching computing in their communities within the context of their culture and language.

  PublisherOA PDFDOI

• Ekko: Fully Decentralized Scheduling for Serverless Edge Computing

  2025-06-03

  article

  While originally designed for the cloud, the benefits of the serverless paradigm are vital in Edge/Fog computing environments. In this paper, we propose Ekko, a novel decentralized edge serverless scheduling system, which enables a large number of serverless applications to run simultaneously at the edge through the Functionas-a-Service (FaaS) model. The key insight is to re-architect the common centralized or hierarchical scheduling systems into a fully decentralized one by using the distributed hash table (DHT) based peer-to-peer (P2P) model, in which many distributed schedulers operate autonomously without any centralized state. In sharp contrast to existing studies, any edge node in our system can act as a scheduler, a function worker, a query forwarder, or a storage node, and flexibly switch between these roles, thereby significantly improving scalability and adaptivity. Ekko introduces three design innovations: a boundary-aware P2P organization, distributed shadow schedulers with a keychain scheduling algorithm, and a distributed locality-aware bucket image store. Our evaluation on 500 Amazon EC2 nodes shows that, compared to the state-of-the-art, Ekko reduces the 90<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-</inf>th percentile tail queue wait time by up to 96.6 %, the scheduling time by up to 38.5 %, and the total deployment time by up to 89.5 %, while efficiently scaling to millions of function invocation requests on thousands of edge nodes.

  PublisherDOI

• WIP: Context-Aware AI in Learning Management Systems for Computer Science Courses

  2025-11-02

  articleSenior author

  The growing adoption of virtual assistants in education has sparked interest in integrating artificial intelligence (AI) into Learning Management Systems (LMSes). This study presents ongoing research on the development and evaluation of an AI-powered plugin for the Moodle LMS that implements a context-aware virtual assistant for Computer Science 1 (CS1) courses. The proposed framework comprises three components: (1) a chat interface embedded within the student's Moodle session, (2) an interchangeable large language model (LLM) hosted locally or via the cloud, and (3) a microservice using LangChain to embed real-time student data in prompts before inference. The main objective is to develop a virtual assistant that seamlessly integrates with the Moodle platform and delivers context-aware, personalized responses. The plugin was built using Moodle's open-source block-plugin framework. A Python-based FastAPI microservice connects the front end to the LLM, utilizing LangChain to access student-specific data-such as grades, submitted code, and chat history-during prompt construction. System performance was assessed using benchmarks for latency, context awareness, and hallucination rates. Preliminary results in a simulated environment show that the system effectively embeds student data to generate accurate responses with low latency. The interchangeable LLM design offers flexibility in deployment configurations for various use cases. This research demonstrates the technical feasibility of embedding AI assistants in LMS environments and sets the stage for future mixed-methods studies. Upcoming pilot implementations will assess the virtual assistant's effectiveness in supporting student engagement and improving learning outcomes in real-world settings.

  PublisherDOI

• AgileDART: An Agile and Scalable Edge Stream Processing Engine

  IEEE Transactions on Mobile Computing · 2025-01-07 · 10 citations

  article

  Edge applications generate a large influx of sensor data on massive scales, and these massive data streams must be processed shortly to derive actionable intelligence. However, traditional data processing systems are not well-suited for these edge applications as they often do not scale well with a large number of concurrent stream queries, do not support low-latency processing under limited edge computing resources, and do not adapt to the level of heterogeneity and dynamicity commonly present in edge computing environments. As such, we present AgileDart, an agile and scalable edge stream processing engine that enables fast stream processing of many concurrently running low-latency edge applications' queries at scale in dynamic, heterogeneous edge environments. The novelty of our work lies in a dynamic dataflow abstraction that leverages distributed hash table-based peer-to-peer overlay networks to autonomously place, chain, and scale stream operators to reduce query latencies, adapt to workload variations, and recover from failures and a bandit-based path planning model that re-plans the data shuffling paths to adapt to unreliable and heterogeneous edge networks. We show that AgileDart outperforms Storm and EdgeWise on query latency and significantly improves scalability and adaptability when processing many real-world edge stream applications' queries.

  PublisherDOI

• Predictive Analysis of Student Dropout in the Computer Science Program at the University of Brasilia

  2025-11-02

  articleSenior author

  Student dropout is a major challenge in Brazilian public higher education, particularly in computer science programs, where rates can exceed 40%. As most of Brazil's top universities are public and tuition-free, high dropout rates not only affect individual academic paths but also strain the broader educational system. This study investigates dropout factors beyond academic performance, incorporating socioeconomic variables, university admission methods, and the effects of the 2012 Quota Law, which reserves 25% of university admission slots for low-income students. This research explores how educational data mining, enhanced by socio-economic data, could identify key predictors of dropout in computer science programs. Using student data collected from a top federal university over the period 2013-2020, the study applies three machine learning algorithms - Gradient Boosting Machine (GBM), Support Vector Machine (SVM), and Random Forest (RF) - to build predictive models. The results show that the models achieved over 90% accuracy in predicting dropout. The most significant predictors include Grade Point Average (GPA), performance in introductory programming courses, and the mode of admission. These findings offer valuable insights for university administrators seeking to reduce dropout rates. While the study focuses on one institution, its implications extend across Brazil's federal university system, where similar admission and evaluation policies are employed. The findings suggest that combining academic and socio-economic data can inform more effective retention strategies in computer science programs.

  PublisherDOI

• Research Experience for Undergraduates for Inclusion of African American and Hispanic Students in Computer Science Majors: A Literature Mapping

  2025-11-02

  articleSenior author

  Research Experience for Undergraduates (REU) programs are designed to support students' progression to graduate education and improve retention in undergraduate majors. In computing fields, African American and Hispanic students remain underrepresented. This study addresses the research question: What does the literature reveal about the impact of REU programs on these populations in computing majors in the U.S.? Five sub-questions guide this work: 1) What activities are associated with REU programs? 2) What is the duration of these programs? 3) What do evaluations report? 4) How do REUs targeting African American and Hispanic students differ from general broadening participation initiatives? 5) How are these programs sponsored? To answer these questions, we conducted a systematic literature review covering the period from 2009 to 2024. Searches were conducted across four databases: ACM Digital Library, IEEE Xplore, Scopus, and Web of Science. After screening, 618 papers were analyzed and 61 were deemed relevant to the topic. The search occurred in five stages between 2019 and 2025, each covering specific publication years. Findings suggest that REU programs contribute positively to students' confidence, sense of belonging, retention in computing majors, and interest in research careers. This mapping offers a comprehensive overview of REU-related efforts over the past 16 years and can inform the design of future initiatives aimed at supporting underrepresented groups in undergraduate research.

  PublisherDOI

• Women in Cybersecurity: A Literature Mapping (2010-2024)

  2025-11-02

  article

  Full Paper. The field of computer science, particularly cybersecurity, remains male-dominated, with women underrepresented in both the workforce and academia. To address this disparity, various global initiatives have aimed to recruit and retain women in cybersecurity. This systematic literature mapping reviews academic efforts to promote women's inclusion in the field over past 15 years (2010-2024). The main research question is: What types of publications address activities aimed at including more women in cybersecurity? This question is divided into four sub-questions: (1) What types of academic publications exist and how are they distributed by year and country? (2) At what educational levels (K-12, undergraduate, graduate) are these activities implemented? (3) What types of initiatives have been implemented? (4) What professional challenges and opportunities are highlighted? Using the Scopus and Web of Science databases, 74 relevant academic publications were identified based on inclusion criteria (academic level, date range, field of study, and focus on women's inclusion). Exclusion criteria ruled out short papers, non-English documents, and those that did not address inclusion activities. The findings show a growing body of literature, including initiatives at all levels of education. Activities range from K-12 gamified programs to undergraduate research engagement. Other studies assess workplace perceptions or analyze gender inclusion strategies through public and private sector documents. This mapping offers a comprehensive overview of global academic efforts to include women in cybersecurity and can inform future initiatives aimed at improving diversity in this critical field.

  PublisherDOI

• ChatGPT as a Teaching Assistant in a CS1 course: An Experience Report

  2025-11-02

  articleSenior author

  This innovative practice full paper presents the findings of an experiment conducted in an introductory programming course (CS1) at the University of Brasilia, Brazil, comparing the effectiveness of undergraduate teaching assistants and ChatGPT. CS1 is often a challenging course for students, typically supported by multiple teaching assistants. With the growing use of generative artificial intelligence (AI) in education, particularly chatbot tools like ChatGPT and Gemini, virtual assistants are playing an increasingly prominent role in programming instruction, even during face-to-face lab sessions. The experiment involved 41 undergraduate students over three consecutive weeks, with one 80 -minute session per week. The students were divided into two groups: one supported by human teaching assistants and the other assisted by ChatGPT. Participants were grouped to balance prior computer science experience and demographic factors. After each lab session, students completed a survey and their submitted work was assessed for accuracy. Additionally, exam scores and students' perceptions of the learning experience were analyzed to evaluate the impact of each method on performance and engagement. The findings showed that students assisted by ChatGPT completed more assignments in less time, with submission accuracy comparable to that of the control group. While students acknowledged the helpfulness of ChatGPT, many still preferred human assistance. This study adds to the growing body of literature on AI in education, which highlights generative AI's ability to support tasks such as coding, debugging, and concept clarification. However, few studies have directly compared AI assistance to human support in classroom settings. These findings provide valuable insights into the potential of generative AI tools like ChatGPT to complement or even replace traditional teaching methods in programming education.

  PublisherDOI

• Totoro: A Scalable Federated Learning Engine for the Edge

  2024-04-18 · 9 citations

  articleOpen access

  Federated Learning (FL) is an emerging distributed machine learning (ML) technique that enables in-situ model training and inference on decentralized edge devices. We propose Totoro, a novel scalable FL engine, that enables massive FL applications to run simultaneously on edge networks. The key insight is to explore a distributed hash table (DHT)-based peer-to-peer (P2P) model to re-architect the centralized FL system design into a fully decentralized one. In contrast to previous studies where many FL applications shared one centralized parameter server, Totoro assigns a dedicated parameter server to each individual application. Any edge node can act as any application's coordinator, aggregator, client selector, worker (participant device), or any combination of the above, thereby radically improving scalability and adaptivity. Totoro introduces three innovations to realize its design: a locality-aware P2P multi-ring structure, a publish/subscribe-based forest abstraction, and a bandit-based exploitation-exploration path planning model. Real-world experiments on 500 Amazon EC2 servers show that Totoro scales gracefully with the number of FL applications and N edge nodes, speeds up the total training time by 1.2 × -14.0×, achieves O (logN) hops for model dissemination and gradient aggregation with millions of nodes, and efficiently adapts to the practical edge networks and churns.

  PublisherDOI

• Programa de Monitoria da Disciplina de Programação Introdutória na Universidade de Brasília

  Revista Brasileira de Informática na Educação · 2024-03-17 · 4 citations

  articleOpen accessSenior author

  De acordo com o relatório da ACM intitulado Retention in computer science undergraduate programs in the US: Data challenges and promising interventions, sobre retenção em cursos de Ciência da Computação, a primeira disciplina de programação, chamada de CS1 (Computer Science 1) no relatório, pode influenciar a permanência do aluno em um curso de computação. Na Universidade de Brasília (UnB), a primeira disciplina de programação tem um alto índice de reprovação. Neste contexto, foi criado um novo Programa de Monitoria para Algoritmos e Programação de Computadores (APC), a primeira disciplina de programação dos cursos de computação na UnB. Este novo programa é composto por atendimentos aos sábados, atendimento individual via agendamento, busca ativa pelos alunos com baixo rendimento, aulões de revisão aos sábados, e acompanhamento em aulas práticas. Neste artigo é apresentado o relato de experiência de quatro semestres do programa, sendo a sua primeira edição no segundo semestre de 2020, durante a pandemia, com ensino remoto, até o primeiro semestre de 2022.1, o retorno ao ensino presencial. Neste artigo, é descrita a metodologia do programa, análise dos resultados desses dois anos de aplicação do programa, e as lições aprendidas.

  PublisherOA PDFDOI

Frequent coauthors

• Maristela Holanda

  Universidade de Brasília

  34 shared

• Robert W. Wisniewski

  17 shared

• Orran Krieger

  16 shared

• Jonathan Appavoo

  14 shared

• Raquel Carneiro Dörr

  Mitchell Institute

  12 shared

• Fernanda Sousa

  Mitchell Institute

  12 shared

• Bryan S. Rosenburg

  9 shared

• Liting Hu

  8 shared

Awards & honors

• ACM Distinguished Speaker, 2009-2014
• ACM Distinguished Scientist, 2011
• IBM Research Division Award, 2011
• Pat Goldberg Best Paper Award, 2009
• Best Paper Award at USENIX File and Storage Technologies (FA…