Hello, I'm

Jiahui An

PhD Candidate at UZH & ETH Zurich

Studying multimodal human interaction with AI systems. I build representations of user state by integrating behavioral traces, language/LLM interaction data, and sensor streams to understand cognitive load, attention, and stress.

View My Research Get In Touch
Jiahui An
Scroll

About Me

My Research Interest

I know BCI often evokes images of moving robotic arms with thought, but I’m genuinely fascinated by something subtler: how our brain whispers its internal state. In my PhD, I explore ways to decode cognitive load, stress, and attention from EEG and biosignals to create AI systems that actually understand how you feel, not just what you type.

My work bridges neuroscience and AI, using spike-based processing 🧠 to capture the temporal dynamics of human cognition. I build Spiking Neural Networks (SNNs) that can run efficiently on neuromorphic chips, enabling wearable devices to monitor mental states in real-time without draining the battery.

What excites me most? Closing the loop with humans! I love designing experiments where the system adapts to the user's workload on the fly. Whether it's an adaptive learning platform or an AI assistant, my goal is to make technology more empathetic and responsive to our natural cognitive rhythms.

Human-AI Interaction

Designing adaptive systems that respond to cognitive load and attention

Multimodal Sensing

Integrating EEG, eye tracking, PPG, EDA, and behavioral data

Neuromorphic Computing

Deploying spiking neural networks on edge hardware

Education

2022 - Present

Joint PhD in Neuroinformatics

University of Zurich & ETH Zurich

Zurich, Switzerland
2019 - 2021

M.S. Neural & Behavioural Sciences

University of Tübingen & IMPRS

Tübingen, Germany
2017 - 2019

B.S. Bioinformatics

University of Tübingen

Tübingen, Germany

Research Projects

Exploring the intersection of human cognition, AI systems, and multimodal sensing

Ongoing Projects

Ongoing Jan 2026 - Present

Multimodal Human-LLM Interaction for Workload-Aware Assistance

Investigating how LLM assistance changes behavior, attention, and cognitive load during coding tasks. Developing sensor-informed interaction metrics for adaptive interfaces.

My Role

Co-Lead Scientist: Concept, study design, implementation, data collection & analysis

What's Innovative

First study to combine physiological sensing with LLM interaction logs to understand real-time cognitive state during AI-assisted coding. Creates foundation for truly adaptive AI assistants that respond to user mental state.

Key Contributions

  • Designed human-subject protocol combining eye tracking + physiological signals with task telemetry
  • Built Python tooling for time-alignment, segmentation, and annotation of multimodal streams
  • Extracted language/interaction features related to self-report and objective performance
EEG Eye Tracking LLM Python Human-AI Interaction
Details available upon request
Active Aug 2025 - Present

Real-Time Neuromorphic Biosignal System for Adaptive Learning

Built a closed-loop neuroadaptive system that adjusts audio playback speed based on inferred cognitive load with real-time streaming acquisition and neuromorphic deployment.

My Role

System Architect & Lead Developer: Concept, full-stack implementation, hardware integration, user studies

What's Innovative

Novel human-in-the-loop co-adaptation paradigm where both the learner and the system adapt together. Ultra-low-power neuromorphic inference enables deployment in real-world wearable scenarios.

Key Contributions

  • Human-in-the-loop co-adaptation with modality-specific update rates and feature-age auditing
  • Streaming EEG/PPG/EDA via LSL with latency-aware buffering
  • Deployed spiking models on Dynap-SE/Dynapse hardware; evaluated accuracy/latency trade-offs
  • Designed experiment protocol to quantify how adaptive policies affect performance over time
Neuromorphic Real-time LSL Dynapse I Closed-loop
Details available upon request

Completed Projects

🏆 Best Paper Nominee Apr - Aug 2025

Edge-AI Infrastructure for Real-time Cognitive State Monitoring

Architected a low-latency inference pipeline for decoding cognitive workload from physiological signals on resource-constrained edge devices for Air Traffic Control applications.

My Role

First Author & Lead Researcher: Algorithm design, neuromorphic deployment, evaluation, paper writing

What's Innovative

First demonstration of cognitive workload classification on mixed-signal neuromorphic hardware (DYNAP-SE) achieving μW power consumption, enabling always-on monitoring for safety-critical applications like air traffic control.

Key Contributions

  • Modeled cognitive workload from EEG + eye-tracking as operator-state signal for shared autonomy
  • Deployed quantized spiking models on DYNAP-SE for μW-range inference
  • Comprehensive latency/power/accuracy trade-off analysis

Outcome

Accepted at ACM/IEEE ICONS 2025, Best Paper Award Nominee

Edge AI DYNAP-SE ATC Dataset Neuromorphic
View Code View Project Details
Oral @ IEEE CIBCB 2025 Nov 2024 - Mar 2025

Multimodal Spiking Neural Network for Workload Monitoring

Developed multimodal fusion architectures to align heterogeneous time-series data into a unified latent representation for robust state estimation.

My Role

First Author: Architecture design, implementation, ablation studies, evaluation

What's Innovative

Novel spiking neural network architecture for multimodal biosignal fusion that handles heterogeneous sampling rates and missing modalities gracefully, key for real-world wearable deployment.

Key Contributions

  • Multimodal fusion of EEG, EDA, Temperature, PPG using PyTorch/snnTorch
  • Systematic ablations on modality contribution and robustness to noisy/missing channels
  • Reproducible evaluation metrics and fusion strategy comparison

Outcome

Accepted (Oral Presentation) at IEEE CIBCB 2025, Tainan, Taiwan

SNN PyTorch Multimodal Fusion snnTorch
View Code View Project Details
Neurophotonics Journal Nov 2022 - Oct 2024

Development of fNIRS Biomarkers of Mental Workload

Demonstrated that short-channel regression improves cortical activation estimates for working memory load assessment using functional near-infrared spectroscopy.

My Role

First Author: Experimental design, data collection, signal processing, statistical analysis, paper writing

What's Innovative

Systematic comparison of fNIRS preprocessing pipelines demonstrating that short-channel regression significantly improves the reliability of cortical activation estimates for cognitive load, establishing best practices for the field.

Key Contributions

  • Modeled N-back working memory effects across multiple load levels
  • Benchmarked preprocessing choices and validated short-channel regression
  • Supervised development of fnirsPy Python wrapper

Outcome

Published in Neurophotonics (2025) — DOI: 10.1117/1.NPh.12.1.015002

fNIRS Working Memory MATLAB Python
View Project Details

Master Thesis: Neural Correlates of Cognitive Impairments

Children Hospital Zurich | Apr - Sep 2021

Linked structural MRI/DTI markers with cognitive outcomes in pediatric focal epilepsy using FSL/FreeSurfer and R.

VR-based Driver Fatigue Assessment

ETH Zurich | Aug - Nov 2020

Designed VR driving-simulator study with eye-tracking to quantify fatigue and attention with safety-relevant recommendations.

Multimodal Analysis of Stress & Synchronicity

MPI for Intelligent Systems | Sep 2019 - Apr 2021

Combined thermal imaging, motion capture, and biofeedback using OpenPose for human behavioral dynamics analysis.

Face Race Categorization & Gaze Behavior

MPI for Biological Cybernetics | Nov 2020 - Apr 2021

Designed behavioral + gaze experiments; modeled attention patterns with statistical analysis.

Publications

Peer-reviewed research contributions

Conference Papers

🏆 Best Paper Nominee

Neuromorphic Deployment of Spiking Neural Networks for Cognitive Load Classification in Air Traffic Control

An, J., Donati, E., Indiveri, G., et al.

ACM International Conference on Neuromorphic Systems (ICONS) 2025

Multimodal Spiking Neural Networks for Mental Workload Classification

An, J., Donati, E., Indiveri, G., et al.

IEEE CIBCB 2025 — Tainan, Taiwan

Journal Articles

Functional near-infrared spectroscopy short-channel regression improves cortical activation estimates of working memory load

An, J., Schönhammer, J., Luft, A., et al.

Neurophotonics, 2025

Who wants to use AI and a digital twin for health? – Evidence on the influence of healthcare utilization behaviour from a representative Swiss population survey

Witt, C., An, J., Christen, M.

BMC Health Services Research, 2025

Under Review

Landmark-based EEG Correlates of Cognitive Load in Virtual Navigation

An, J., Cheng, B., Donati, E., Fabrikant, S., et al.

Submitted to IEEE EMBC 2026

View Full Publication List on Google Scholar

Skills & Expertise

Machine Learning

PyTorch snnTorch TensorFlow scikit-learn Multimodal Fusion

Programming

Python MATLAB R Git

Signal Processing

EEG fNIRS PPG/EDA Eye Tracking Time-series Analysis

Research Methods

Study Design Hypothesis Testing Mixed Methods Statistical Modeling

Real-time Systems

Lab Streaming Layer Low-latency Pipelines Multiprocessing Live Dashboards

Hardware

DYNAP-SE Jetson Nano Neuromorphic Chips Edge Devices

Languages

Chinese Native
English Fluent
German Fluent

Beyond Research

The interests and activities that shape who I am outside the lab

Music & Piano

Piano has been a lifelong companion—I find that practicing music exercises a different kind of focus and creativity that complements scientific thinking. Whether learning classical pieces or improvising, it's my way of unwinding after intense research sessions.

Hiking & Nature

Living in Switzerland offers incredible opportunities for alpine adventures. From day hikes in the Uetliberg to multi-day treks in the Alps, being in nature provides perspective and inspiration that often sparks new research ideas.

Travel & Cultural Exchange

Growing up in China, studying in Germany, and now working in Switzerland has given me a deep appreciation for cross-cultural collaboration. I love exploring new places and connecting with researchers from diverse backgrounds.

Community & Mentorship

I'm passionate about supporting the next generation of researchers. I supervise student projects, contribute to tutorials, and actively participate in scientific communities through conference organization and peer review.

VR Gaming & Technology

As someone researching human-computer interaction, I'm fascinated by immersive technologies. VR gaming isn't just fun—it's a way to experience firsthand the interfaces and interactions I study professionally.

Vintage Curation

I have a creative side that manifests in curating vintage items and aesthetics. Finding and restoring unique pieces connects me to history and design in ways that feel refreshingly different from my technical work.

Get In Touch

Interested in collaboration? Let's connect!

Affiliation

Institute of Neuroinformatics
University of Zurich & ETH Zurich

Location

Zurich, Switzerland

Google Scholar LinkedIn GitLab