Gustavo Penha

Research Scientist @Spotify · Working with IR, RecSys, NLP · PhD from @tudelft · ex @AmazonScience ·

[Gustavo Penha @_Guz_]

We wrote a post summarizing our #RecSys2024 paper on bridging search and recommendation with generative retrieval 🧵 (1/N)
https://t.co/vmG7iLr1Em
w. @AliVardasbi, @denadai2, @enricopalumbo91, Hugues Bouchard

[Kamil Ciosek @MLciosek]

For anyone worried their LLM might be making stuff up, we made a budget‐friendly truth serum (semantic entropy + Bayesian). See for yourself: https://t.co/gq8oFP5Eqr
Paper: https://t.co/s6SuQW4sAM

[Aixin Sun @AixinSG]

I doubt to what extent improvements on these datasets would translate to improvements in today's real-world recommendation settings. Reference: https://t.co/YPcJdfMy2p

[Gustavo Penha @_Guz_]

@svakulenk0 Thanks! But I won’t go this time :(
my co-authors will present this time

[Gustavo Penha @_Guz_]

Happy to share our #recsys25 paper: “Evaluating Podcast Recommendations with Profile-Aware LLM-as-a-Judge”.

🧠 90 days of listening → natural-language user profiles → LLM judges alignment
📊 Aligns with human eval.

With amazing Spotify co-authors.
📄 https://t.co/JpdOLQ3yRs

[Gustavo Penha @_Guz_]

@svakulenk0 Yes! We calculate the agreement between the llm judge and users in the paper

[Gustavo Penha @_Guz_]

📄 https://t.co/w2PnNxEn0J

[Gustavo Penha @_Guz_]

Excited to share our paper “Semantic IDs for Joint Generative Search & Recommendation” @ RecSys'25

🧠 Jointly fine-tuning embeddings for both tasks → shared Semantic IDs that work for search and recs ⚖️
📦 No more task-specific trade-offs!

[Sumit @_reachsumit]

Semantic IDs for Joint Generative Search and Recommendation

@_Guz_ et al. at Spotify introduce a bi-encoder model fine-tuned on both search and recommendation tasks to obtain item embeddings, followed by construction of unified Semantic ID space.

📝https://t.co/mTERH16wib

[Sumit @_reachsumit]

Evaluating Podcast Recommendations with Profile-Aware LLM-as-a-Judge

Spotify introduces a profile-aware LLM framework for evaluating personalized podcast recommendations using natural-language user profiles distilled from listening history.

📝https://t.co/Rk8qUS0V2P

[Sumit @_reachsumit]

Describe What You See with Multimodal Large Language Models to Enhance Video Recommendations

@denadai2 et al. at Spotify use multimodal LLMs to generate natural-language descriptions of video content for better recommendations

📝https://t.co/zloHLosVzd
👨🏽‍💻https://t.co/PlWzE5TRWp

[Marco De Nadai @denadai2]

What if we could use off-the-shelf Multimodal Large Language Model to enrich current video recommendation models?

This is what we asked ourselves in our recent #recsys2025 paper https://t.co/1dqzYgM8LR
🧵

[Gustavo Penha @_Guz_]

🔗 Blog Post: https://t.co/yfuAGZSySj
📚 Paper: https://t.co/ommhSyyRAx

[Gustavo Penha @_Guz_]

🔎 LLM alignment techniques can enhance query expansion by eliminating the need for multiple generations followed by re-ranking/filtering steps.

Check out this work led by @adam_x_yang during his internship with us at @SpotifyResearch w. @enricopalumbo91 and Hugues Bouchard⬇️

[Sumit @_reachsumit]

Adaptive Repetition for Mitigating Position Bias in LLM-Based Ranking

Spotify introduces a dynamic early-stopping method that adaptively determines repetitions needed for each ranking instance, reducing LLM calls by 81% while preserving accuracy.

📝https://t.co/TjcyF0RvlG

[Sumit @_reachsumit]

Aligned Query Expansion: Efficient Query Expansion for Information Retrieval through LLM Alignment

@adam_x_yang et al. leverage LLM alignment techniques to fine-tune models for generating query expansions that directly optimize retrieval effectiveness.

📝https://t.co/sAnWpP67wq

[Sumit @_reachsumit]

Contextualizing Spotify's Audiobook List Recommendations with Descriptive Shelves

Spotify introduces a pipeline that generates personalized audiobook recommendations with descriptive shelves to help users explore content based on their interests.

📝https://t.co/dzJvBkOzZI

[Gustavo Penha @_Guz_]

Blog post: https://t.co/0wCOjOPCnL
Paper: https://t.co/yEsn2JxMTu

[Gustavo Penha @_Guz_]

We just published this blog post about our research on music track search with generative retrieval. 🧵

With @enricopalumbo91 @adamianou @peputo Timothy Christopher, Alice Wang, Hugues Bouchard, @mounialalmas

[Gustavo Penha @_Guz_]

The best-performing ID strategy was to use collaborative-filtering embeddings as input to the discretization approach for semantic IDs