Scholar
Steven Abreu
Google Scholar ID: CqbIOvMAAAAJ
University of Groningen (Previously: Intel Labs, Google)
Citations & Impact
Citations
429
H-index
9
i10-index
8
Publications
20
Co-authors
9
list available
Publications
Resume (English only)
Academic Achievements
- May 2025: Presented “Large Language Models on a Tiny Power Budget” at ISCAS 2025
- Apr 2025: Presented four workshop papers at ICLR 2025 in Singapore
- Dec 2024: Co-authored paper “Steering Large Language Models using Conceptors” presented at the MINT Workshop, NeurIPS 2024
- Jul 2024: Presented work on quantized SSMs at ICML 2024 (Vienna) and neuromorphic programming at ICONS 2024 (Virginia)
- Jun 2023: Presented neuromorphic cytometry paper at CVPR 2023 (event-based vision workshop, Vancouver)
- Oct 2023: Poster presentation on neuromorphic programming at NNPC 2023 (Hannover)
- Apr 2023: Gave a talk on bio-inspired hardware interfaces at CHI 2023 (Hamburg)
Research Experience
- Visiting researcher at ETH Zurich and University of Gent during PhD studies
- Research intern at Google (Waterloo, Canada), working on efficient and adaptive AR user interfaces with multimodal LLMs
- Research intern at Intel’s Neuromorphic Computing Lab, developing neuromorphic transformer-like architectures based on state space models
- Three-month research stay at the Institute of Neuroinformatics (ETH/UZH, Zurich) in 2022, focusing on mixed-signal neuromorphic hardware
- Three-month research stay at the Photonics Research Group, University of Gent (2022), working on photonic computing and neuromorphic flow cytometry
- Multiple participations in the Telluride Neuromorphic and Cognitive Computing Summer Research Program
Background
- PhD student in the AI Department at the University of Groningen, working in the MINDS group under Prof. Herbert Jaeger
- Affiliated with CogniGron and partially funded by Post-Digital
- Maintains a dual research focus on advancing machine learning and exploring brain-inspired computing paradigms and AI hardware
- Enjoys interdisciplinary research spanning AI, computer science, mathematics, neuroscience, physics, and cognitive science
- Within ML, interested in efficiency (e.g., model compression, low-power hardware), continual learning (e.g., brain-inspired neuromodulation or plasticity rules), meta-learning, and automated machine learning
- Works on representation engineering and mechanistic interpretability of large language models to advance aligned and interpretable AI
- Researches physical and brain-inspired computing, developing theories that align computation with physics to better utilize neuromorphic chips, photonic devices, and other physical computing systems
- Aims to develop computational abstractions in physical substrates (e.g., via the Neuromorphic Intermediate Representation, NIR), hardware-compatible efficient learning algorithms, and principled methods for programming novel AI hardware
