Pipeline Unificado para Tomada de Decisão Financeira com Agentes Multimodais e Descentralizados

VISION is a framework for robust and interpretable code vulnerability detection using counterfactual data augmentation. It leverages GNNs, LLM-generated counterfactuals, and graph-based explainability to mitigate spurious correlations and improve generalization on real-world vulnerabilities (CWE-20).

Benchmarking a novel bio-informed RNN architecture under Dr. Cheng Zhang at KCL

a clean, educational agent framework for learning and prototyping

My school project (MSc Applied AI): Brain tumour classification with pre-trained models and a custom CNN

Análise Aprimorada de Intervenção para Ansiedade com LLM Fine-Tuned

Reproducible NLP research repo for extremist vs non-extremist social-media text classification.

🏎️ Predict Formula 1 podium finishers with 96.6% accuracy using machine learning and extensive telemetry data for insightful race forecasts.

Method to generate counterfactuals using dichotomous (Binary search) logic

Transformer-based models for stereotype detection in text with explainability via LIME, SHAP, and attention visualizations.

Explainability and Interpretability: Understanding and Implementing AI Explainability

The independent reimplementation; the fusion experiment (feature-level vs decision-level) and its negative result; the second-dataset (PaySim) generalization test; the systematic SHAP/LIME interpretability analysis; the cost/threshold framing; the serving system, explainability API, and demo; and the engineering conclusions drawn from all of it.

explainability of models applied to obstetric medical imaging

Exploring privacy-explainability trade-offs using differential privacy, federated learning, and membership inference attacks

An implementation of Text2Concept paper