Technical Resume
Machine Learning Specialist and NeuroAI Researcher
Machine learning specialist with 3+ years of experience building deep learning systems for neural signal processing, multimodal biomedical data, and reproducible research workflows. I work across EEG, fMRI, behavioral data, and large-scale compute environments to build robust, responsible AI for neuroscience and healthcare.
Core Themes
Self-supervised learning, multimodal modeling, signal processing, neurotechnology, open science, and practical ML systems that remain reproducible under research constraints.
Professional Summary
I specialize in building scalable deep learning pipelines for neuroscience and biomedical applications, with hands-on experience in EEG, fMRI, behavioral data, and multimodal integration. My work combines model development, reproducible data handling, and high-performance compute workflows, with a strong interest in foundation models, responsible AI, and clinically relevant research systems.
Experience
Graduate Research Assistant, NeuroAI
- Built deep learning models in TensorFlow and PyTorch, improving diagnostic accuracy by 35% in neuroimaging analysis workflows.
- Developed a foundation-model pretraining pipeline integrating EEG, fMRI, and behavioral signals across 3000+ subjects.
- Reduced manual data handling time by 40% through pipeline automation in Python and MATLAB.
- Optimized Slurm-based training and data workflows on Compute Canada, spanning 8+ GPU nodes and roughly 3TB data throughput.
- Maintained reproducible workflows with Git while contributing to open-access datasets and peer-reviewed research outputs.
Python Developer, AI/ML Team
- Designed and deployed deep learning models for video and audio analysis, reaching 95% detection accuracy.
- Reduced model latency by 40% through optimization and streamlined inference design.
- Improved deployment speed by 25% by tightening Git and Docker based CI/CD workflows.
Biomedical Engineer, AI for Portable Diagnostics
- Designed machine learning algorithms for portable ultrasound devices, improving detection accuracy by 35% while meeting regulatory benchmarks.
- Led FDA-aligned performance validation, including A/B testing and statistical reporting.
- Contributed to 2 grant proposals and technical packages that supported clinical pilot testing with 3 hospital partners.
- Worked across engineering, clinical, and software teams in a university-affiliated health innovation environment.
Selected Projects
Multimodal Modeling of ADHD in Youth
Built predictive models integrating fMRI connectivity and behavioral features from 3000+ subjects, using time-series extraction, graph-based metrics, and Python-based signal pipelines to study sex-based disparities in ADHD diagnosis.
Machines That Imagine: VAE and Diffusion Models
Developed VAE, diffusion, and SimSiam workflows on SVHN and CIFAR-10, focusing on image synthesis, latent space learning, and reproducible experimentation.
Transformers vs. RNNs on Large-Scale Review Data
Compared GRUs and Transformer models on 35M Amazon reviews to analyze performance, bias, and attention behavior in large-scale text classification.
Decoding Sentiment at Scale
Classified 1M+ text samples using ensemble ML and NLP pipelines, integrating LIME to improve interpretability and stakeholder trust.
Publications
Towards Multi-Brain Decoding in Autism: A Self-Supervised Learning Approach
Applied self-supervised deep learning to EEG hyperscanning data, achieving 78.13% classification accuracy and demonstrating how SSL can improve diagnostic performance in low-label clinical settings.
A Neurodynamic Model of Inter-brain Coupling in the Gamma Band
Modeled gamma-band inter-brain coupling with computational neural dynamics to study social interaction and multi-subject EEG phenomena.