Hello, I'm

Arturo Fredes

ML Engineer & Researcher

I'm a physicist and mathematician working at the intersection of AI research and real-world application. My background spans both academia and industry, which has shaped my approach: rigorous in method, pragmatic in execution. My work focuses on machine learning and large language models, from evaluation and benchmarking to applied research and deployment. I'm particularly drawn to interpretability and causal reasoning, areas where I've contributed both in research settings and through published work. I believe the most valuable work happens where fundamental research meets practical impact, and that's the space I aim to occupy.

GitHub LinkedIn Scholar ORCID

Research Interests

LLMsNLPFoundation ModelsBenchmarkingExplainabilityComputer Vision Deep Learning

Languages

Spanish · Native

English · C2

German · B2

Publications & Theses

KDD 2024 Workshop · MSc Thesis — Universitat de Barcelona

Using LLMs for Explaining Sets of Counterfactual Examples to Final Users

Arturo Fredes et al.

Study of applications of LLMs in the business domain, focusing on using large language models to generate natural language explanations of sets of counterfactual examples, making causal inference results more interpretable for end users. This work was developed as a Master's thesis and later presented as a conference paper at KDD 2024 Workshop — Causal Inference and Machine Learning in Practice.

Paper Thesis Workshop Code

Causal Inference Counterfactuals XAI LLMs

Mathematics Thesis (Honours) — Universidad de Zaragoza

Recurrent Neural Networks for Time Series Forecasting in Complex Dynamical Systems and its Applications

Arturo Fredes — Collaboration with Editorial Edelvives

Application of recurrent neural networks (RNN/LSTM) to time series forecasting in complex dynamical systems, with practical applications developed in collaboration with Editorial Edelvives for demand forecasting.

Thesis Code

RNNs Dynamical Systems Time Series Nonlinear Sys.

Physics Thesis — Universidad de Zaragoza

Design and Manufacture of Superconducting Radiation Detectors

Arturo Fredes

Design and manufacturing process of superconducting radiation detectors, exploring the physics of superconducting materials and their applications in particle detection.

Thesis Code

Superconductivity Quantum Materials Exp. Physics

Work Experience

Feb 2026 — Current

Senior Machine Learning Engineer

Multiverse Computing (Evaluations Team)

Team Leadership

Led a team of 4 Machine Learning Engineers, coordinating research, development and delivery of LLM evaluation projects.

LeadershipAgile

LLM Benchmarks & Internal Models

Research and implementation of industry-leading benchmarks for the evaluation of LLMs across modalities. Benchmarking and evaluation of internal models.

LLMsBenchmarksModel Cards

Hypernova 60B 2605 — Artificial Analysis

Collaboration with Artificial Analysis to add Hypernova 60B 2605 into their leaderboards, aligning evaluation methodologies.

Article Artificial Analysis

Pulsar 16B — NVIDIA Validation

Collaboration with NVIDIA to align evaluation methodologies. NVIDIA validated our internal evaluations of Pulsar 16B, the compressed version of Nemotron 3 Nano 30B A3B.

Article

Mar 2025 — Feb 2026

Technical Researcher

Instituto de Nanociencia y Materiales de Aragón

Project: IASOMM — Inteligencia Artificial para el desarrollo de tecnología de materiales

TEM Image Denoising

Research and implementation of AI algorithms for denoising transmission electron microscopy images, using classical methods in the transform domain and CNNs with state-of-the-art computer vision techniques.

Computer VisionCNNsUnsupervised Learning Fourier Transform

On-Premise RAG System

End-to-end development of an on-premise agent-based RAG system for scientific production management, including embedding model fine-tuning, retrieval and reasoning logic, and full UI implementation.

LLMsRAGAgentsFine-tuning

Nov 2023 — Aug 2024

Assistant Researcher

Universitat de Barcelona

LLMs for Explainability & Causality

Study of applications of LLMs in the business domain, including RAG-based models and use cases in interpretability and causality.

LLMsRAGXAI

KDD 2024 Workshop Paper

This work was extended into my Master's thesis and later presented as a conference paper at KDD 2024 Workshop — Causal Inference and Machine Learning in Practice.

Causal InferenceCounterfactualsXAILLMs