Profile photo of Rafael Ferreira, AI Researcher at NOVA School of Science and Technology

Rafael Ferreira

Ph.D. Student at NOVA School of Science and Technology, Portugal
rah.ferreira [at] campus.fct.unl.pt

About Me

Hello!

I'm Rafael Ferreira, a Ph.D. student at NOVA School of Science and Technology, Portugal. My research focuses on AI-based conversational assistants for complex manual tasks, aiming to improve the robustness and human-like behavior of conversational agents.

I also had the opportunity to lead the award-winning TWIZ team in both editions of the Alexa Prize TaskBot Challenge.

My interests include Conversational Agents, Natural Language Processing, Machine Learning, and Multimodal AI.

Academic Degrees

Ph.D. Student
NOVA School of Science and Technology logoNova School of Science and Technology
Exploring AI-based conversational assistants and user modeling for complex manual tasks.
Supervisors: Prof. João Magalhães, Prof. David Semedo.
2021 - Present
M.Sc. in Computer Science
NOVA School of Science and Technology logoNova School of Science and Technology
Thesis: Context Tracking in Conversational Search: from Utterances to Neural Embeddings. [PDF] [Cite]
Supervisors: Prof. João Magalhães, Prof. David Semedo.
2018 - 2021
B.Sc. in Computer Science
NOVA School of Science and Technology logoNova School of Science and Technology
2015 - 2018

Professional and Research Experience

Amazon logo, Applied Scientist Intern position
Applied Scientist Intern
Agentic AI research to solve real-world, challenging problems.
Seattle, Washington, United States
2025
AI, Data Science, and ML Instructor
CMU Portugal Academy logo, Data Science Instructor positionCMU Portugal Academy
Responsible for the Foundations of Data Science module, focusing on the essential principles and applications. Developed and presented both theoretical and practical classes.
2024

Publications

For a full list, visit my Google Scholar profile.

RedTWIZ: Diverse LLM Red Teaming via Adaptive Attack Planning
Artur Horal, Daniel Pina, Henrique Paz, Iago Paulo, João Soares, Rafael Ferreira, Diogo Tavares, Diogo Glória-Silva, João Magalhães, David Semedo
Amazon NOVA AI Challenge Proceedings, 2025. [PDF] [Cite]
2025
Multi-trait User Simulation with Adaptive Decoding for Conversational Task Assistants
Rafael Ferreira, David Semedo, João Magalhães.
EMNLP, 2024. [PDF] [Cite]
2024
Plan-Grounded Large Language Models for Dual Goal Conversational Settings
Diogo Silva, Rafael Ferreira, Diogo Tavares, David Semedo, João Magalhães.
EACL, 2024. [PDF] [Cite]
2024
TWIZ-v2: The Wizard of Multimodal Conversational-stimulus
Rafael Ferreira et al.
Alexa Prize TaskBot Challenge 2 Proceedings, 2023. [PDF] [Cite]
2023
Grounded Complex Task Segmentation for Conversational Assistants
Rafael Ferreira, David Semedo, João Magalhães.
SIGDIAL, 2023. [PDF] [Cite]
2023
Rating Prediction in Conversational Task Assistants with Behavioral and Conversational-Flow Features
Rafael Ferreira, David Semedo, João Magalhães.
SIGIR, 2023. [PDF] [Cite]
2023
TWIZ: A Conversational Task Wizard with Multimodal Curiosity-Exploration
Rafael Ferreira, Diogo Tavares, Diogo Silva, Frederico Vicente, Mariana Bonito, Gustavo Gonçalves, Rui Margarido, Paula Figueiredo, Helder Rodrigues, David Semedo, João Magalhães.
Alexa Prize TaskBot Challenge Proceedings, 2022. [PDF] [Cite]
2022
Open-domain conversational search assistants: the Transformer is all you need
Rafael Ferreira, Mariana Leite, David Semedo, João Magalhães.
Information Retrieval Journal, 2022. [PDF] [Cite]
2022
Open-Domain Conversational Search Assistant with Transformers
Rafael Ferreira, Mariana Leite, David Semedo, João Magalhães.
ECIR, 2021. Top-8 best papers. [PDF] [Cite]
2021
iFetch: Multimodal Conversational Agents for the Online Fashion Marketplace
Ricardo Gamelas Sousa, Pedro Miguel Ferreira, Pedro Moreira Costa, Pedro Azevedo, João Paulo Costeira, Carlos Santiago, João Magalhães, David Semedo, Rafael Ferreira, Alexander I Rudnicky, Alexander Georg Hauptmann.
ACM Multimedia, MuCAI Workshop, 2021. [PDF] [Cite]
2021
NOVA at TREC 2020 Conversational Assistance Track
Rafael Ferreira, David Semedo, João Magalhães.
TREC CAsT 2020. [PDF] [Cite]
2020

Projects

Team member of RedTWIZ (Red-Teaming WIZard), 2nd place in the Amazon NOVA AI Challenge. Developed red-teaming strategies for safer generative AI.
2024 - 2025
Project Manager for Silicium Sailor, an advanced on-premises Conversational RAG system for Navy/NATO protocols, showcased at REPMUS 2024.
2024
Team Lead of TWIZ, winner of the 2nd Alexa Prize TaskBot Challenge. Developed multimodal conversational AI for complex tasks.
2022 - 2023
Multimodal conversational agents for online marketplaces, in collaboration with Farfetch and academia.
2021 - 2023
Team Lead of TWIZ, 2nd place in the 1st Alexa Prize TaskBot Challenge.
2021 - 2022
NOVASearch
Member of NOVASearch research group, focused on information retrieval, NLP, and multimodal applications.
2019 - Present
GoLocal
Project on context-aware recommendation and global data streams (CMUP-ERI/TIC/0046/2014).
2019 - 2020

Honours and Awards

2nd place, Amazon NOVA AI Challenge 2024
Team RedTWIZ secured 2nd place in the inaugural Amazon NOVA AI Challenge.
2025
1st place, Alexa Prize TaskBot Challenge 2022
Team Lead of TWIZ, 1st place in the 2nd edition.
2023
2nd place, Alexa Prize TaskBot Challenge 2021
Team Lead of TWIZ, 2nd place in the 1st edition.
2022
PhD Scholarship
Awarded an FCT PhD scholarship.
2021 - 2025
Academic Merit Scholarship
Awarded for being among the top-3 highest average grades in the 5th year.
2020

Volunteering

Responsible for instructing, managing, and distributing a team of over 50 volunteers across the conference.
2022
Organizer for ESSIR 2022
Local organizer at the 13th European Summer School in Information Retrieval (ESSIR 2022).
2022
Organizer for Tutorial on Open-Task Multimodal Conversational Assistants presented at both ACM International Conference on Multimodal Interaction (ICMI) and ACM International Conference on Intelligent Virtual Agents (IVA) 2022.
2022
Reviewer
Reviewer for leading international conferences in Computer Science, NLP, and AI. Evaluating the quality of state-of-the-art research.
2021 - Present