Pronounced GAH-zah-leh

Hi, I'm Ghazaleh.

I'm a Machine Learning Specialist and software developer with a background in biomedical engineering and computational neuroscience.

My work sits at the intersection of machine learning research and software development. I enjoy building systems that bring research ideas into practice—from neural-network implementations and scientific data pipelines to agentic workflows, evaluation tools, and GPU-accelerated computing.

I'm currently affiliated with Mila – Quebec AI Institute, working within IVADO's AI and Neuroscience research cluster (R1), where I contribute to software and computational infrastructure for multimodal foundation-model research.

Machine Learning Software Development Agentic AI Scientific Computing

What I Build

My work spans several interconnected areas:

Machine Learning & Deep Learning

Implementing and evaluating deep-learning architectures, developing representation-learning workflows, and working with multimodal data.

Agentic AI & Evaluation

Developing LLM-integrated workflows for data inspection, structured planning, validation, and reproducible evaluation.

Software Development & ML Infrastructure

Designing reusable Python components, developing tests and interfaces, integrating inference systems, and building computational workflows for GPU/HPC environments.

Scientific & Multimodal Computing

Working with heterogeneous datasets, including EEG, fMRI, video, and behavioral recordings, where data quality, temporal alignment, and provenance are essential.

From Research to Software Systems

My academic background is in biomedical engineering and computational medicine. My earlier research explored neural dynamics, EEG hyperscanning, and self-supervised learning for multi-brain decoding.

That experience shaped how I approach machine learning: understanding the data, validating assumptions, and developing methods that remain scientifically meaningful.

More recently, my work has expanded toward ML software infrastructure and agentic AI systems.

I am a co-first author of NeuroBuilder: Auditable Agentic Data Curation for Neuro-Foundation Models, accepted at SMASH 2026. The work investigates how agentic reasoning can be combined with deterministic validation, provenance tracking, and researcher oversight to make scientific data adaptation more reliable.

I am also a first author of Towards Multi-Brain Decoding in Autism: A Self-Supervised Learning Approach, published in Neuroinformatics in 2026.

How I Work

I value clear software architecture, careful validation, reproducible experiments, and collaborative development.

I enjoy working across research and engineering teams, translating complex requirements into practical software, and making technical decisions that can be understood, tested, and maintained.

Technical Interests

My broader interests include:

  • Reliable AI systems
  • Scientific foundation models
  • Open-source software
  • Model evaluation
  • Scalable ML infrastructure

Let's Connect

I'm interested in collaborations and opportunities involving machine learning, AI research, software development, and scientific computing.

You can reach me through the private contact form below.

This form lets you contact me without my email address being displayed publicly.