Gustavo Penha

I am a research scientist at Spotify, with a PhD from TU Delft. My work spans information retrieval, recommender systems, natural language processing, and machine learning. I am passionate about building systems that connect people with content (and creators) by understanding their information needs and preferences. I am interested in the entire information-seeking process and in designing and evaluating systems that support it. I am also a photographer in my spare time, you can see my projects here.

selected publications

  1. RecSys workshop paper
    The Disconnect Between Better Descriptive Reasoning Trace Quality and Recommendation Effectiveness
    Penha, Gustavo, Elenter, Juan, Hauff, Claudia, Bouchard, Hugues, Bennett, Paul, and Lalmas, Mounia
    PDF
  2. RecSys workshop paper
    From IR to RecSys: Evaluating LLM-based Judges in Cranfield-style Recommendation Collections
    Penha, Gustavo, Petrov, Aleksandr V., Hauff, Claudia, Palumbo, Enrico, Vardasbi, Ali, D’Amico, Edoardo, Fabbri, Francesco, Wang, Alice, Chandar, Praveen, Lindstrom, Henrik, Bouchard, Hugues, and Lalmas, Mounia
    PDF
  3. RecSys LBR track
    Semantic IDs for Joint Generative Search and Recommendation
    Penha, Gustavo, D’Amico, Edoardo, De Nadai, Marco, Palumbo, Enrico, Tamborrino, Alexandre, Vardasbi, Ali, Lefarov, Max, Lin, Shawn, Heath, Timothy, Fabbri, Francesco, and others,
    PDF
  4. RecSys
    Bridging Search and Recommendation in Generative Retrieval: Does One Task Help the Other?
    Penha, Gustavo, Vardasbi, Ali, Palumbo, Enrico, Nadai, Marco, and Bouchard, Hugues
    PDF Blog Slides

recent presentations

Slides for talk on generative search and recs at Spotify.