Graph neural networks
Designing and analyzing GNNs with attention to expressivity, message passing limits, and useful structural priors
PhD Researcher - Graph Neural Networks - AI for Biological Data
Researching graph structure for more capable AI systems
I am a PhD researcher in Bioinformatics at the Barcelona Supercomputing Center, working on graph neural networks, graph structure, topological rewiring, and AI methods for biological data. My work focuses on how structural inductive biases can improve learning, robustness, and long-range information flow in modern AI models.
- Current role
- PhD Researcher in Bioinformatics at BSC
- Research themes
- GNNs, graph rewiring, topology, biological AI
- Base
- Barcelona, Spain
Current work spans graph neural networks, graph structure interventions, and AI models for complex biological signals
About
Graph learning research with direct AI implications
My research sits at the intersection of graph representation learning, structural priors, and applied AI for scientific problems. I am particularly interested in how graph topology shapes model behavior, including information bottlenecks, over-squashing, and the inductive biases that make graph-based models more effective.
Alongside graph learning, I work on AI methods for biological data, including recent work connected to RNA foundational models. I care about methods that are both theoretically motivated and useful in real modeling pipelines.
Research Focus
Four directions that define my current work
Graph structure learning
Studying how topology, bottlenecks, and connectivity patterns affect model behavior and representation quality
Topological rewiring
Exploring rewiring strategies that improve long-range information propagation while controlling oversmoothing and graph distortion
AI for biological data
Applying modern representation learning methods to biological sequences and related scientific data problems
Publications
Selected papers across graph learning and biological foundation models
ICLR 2026 Workshop GRaM · Poster
Effective Resistance Rewiring: A Simple Topological Correction for Over-Squashing
Research on using effective resistance as a global signal for graph rewiring to relieve structural bottlenecks in GNNs
arXiv - 2024 · ICLR Workshops 2025
Character-level Tokenizations as Powerful Inductive Biases for RNA Foundational Models
Work on RNA foundation models showing how character-level tokenization can be an effective inductive bias for biological sequences
arXiv - 2024 · DMLR - 2025
TopoBenchmark: A Framework for Benchmarking Topological Deep Learning
Contribution to an open framework for evaluating methods in topological deep learning across standardized benchmarks
ICML Topological Deep Learning Challenge - 2024
ICML Topological Deep Learning Challenge 2024: Beyond the Graph Domain
Workshop paper connected to benchmarking and evaluation beyond standard graph settings in topological deep learning
Blog
Research notes, graph ML intuition, opinions on my own
ICLR GRaM Poster - 2026
Effective Resistance Rewiring
A future long-form post on over-squashing, effective resistance, rewiring trade-offs, and what the paper changes in practice
GRaM Blog - 2026
Graph directionality matters
A blog post on why edge directionality is not a cosmetic detail in graph learning, and how it changes the behavior of message passing models
GRaM Blog - 2026
Fewer edges
A post exploring when simplifying graph connectivity can improve learning by changing information flow, structure, and inductive bias
Technical Stack
Tools behind the research workflow
Graph ML
Graph neural networks, topology-aware learning, rewiring methods, graph benchmarks
Deep Learning
PyTorch, representation learning, foundation models, sequence modeling
Scientific AI
Bioinformatics, biological sequences, RNA modeling, data-driven research workflows
Research Practice
Experiment design, benchmark analysis, technical writing, reproducible model evaluation
Contact
Open to research conversations, collaboration, and graph ML discussions
Email is the fastest way to reach me. You can also follow my publications and online profiles through Google Scholar, GitHub, and LinkedIn.