Electrical Engineering Research Assistant

Tahereh Vasei

•************@*****.*** •Google Scholar•LinkedIn•551-***-**** • New York

Education

New York Institute of Technology, New York, USA

• Ph.D. in Electrical Engineering, Electronic

University of Tehran, Tehran, Iran

• M.Sc. in Electrical Engineering, Digital Electronic Jan 2022 University of Kashan, Isfahan, Iran

• B.Sc. in Electrical Engineering, Electronic 2018 Research Experience

2023 - Present Research Assistant at AI-Driven NeuroScience Lab USA

• Leading PhD research in AI for Neuroscience and Medical Applications, developing deep learn- ing models for biomedical signal classification and mental health issues.

• Focused on combining Transformers, Graph Neural Networks, and CNNs for enhanced accuracy in healthcare data analysis.

• Developed and validated deep learning models, achieving over 90% accuracy for classifying major depression and healthy control brain signals on publicly available datasets.

• Successfully transitioned these models to real-world conditions using real data, achieving over 80% accuracy through strategic model modifications and feature engineering. 2018 - 2022 Research Assistant at Computer-Aided Design (CAD) and TLM Labs Iran

• Developed an efficient and accurate embedded motor imagery-based Brain-Computer Interface

(MI-BCI) optimized for wearable and real-time health applications.

• Leveraged BCI transducer algorithms (IIR filters, common spatial patterns, SVMs) for time- series brain signal preprocessing, feature extraction, and classification, achieving 77% accuracy with hardware implementation.

• Designed the system at the Register Transfer Level (RTL) for ASIC implementation, significantly reducing power consumption (4 mW), latency, and area (0.25 mm ) compared to existing architectures, demonstrating superior performance on a Virtex-7 FPGA. PhD Projects and Innovations

• Generative Model Development (VAE, CVAE, VQ-VAE): Developed advanced Generative AI models

(Variational Autoencoders - VAE, CVAE, VQ-VAE) for high-fidelity image synthesis, demonstrating expertise in deep learning architecture design and self-supervised learning.

• Natural Language Processing

– Sentiment Analysis of Persian Text: Developed LSTM-CNN models on Persian Twitter data, opti- mizing with a greedy search algorithm to enhance sentiment detection accuracy.

– Fake News Detection: Fine-tuned BERT models for COVID-19 fake news detection, integrating advanced neural network architectures for improved reliability.

• Computer Vision

– Image Classification with Vision Transformers: Investigated Vision Transformers and CNNs, achiev- ing high accuracy on the CIFAR-10 dataset through model fine-tuning.

– Image Segmentation with UNet and UNet-TA: Implemented segmentation models on the SUIM dataset, enhancing performance with data augmentation techniques.

• Reinforcement Learning for Fixed-Point Hardware Optimization

– Developed a custom OpenAI Gym environment to train an intelligent agent for automated fixed- point adder design.

– Leveraged RL to navigate complex design trade-offs, optimizing for both hardware area and nu- merical precision, demonstrating proficiency in AI-driven hardware co-design.

• Prefetching in 3-D Spatial Algorithms

– Designed and implemented a prefetcher using SystemVerilog for 3-D spatial algorithms, enhancing computational efficiency in complex environments.

Selected Certifications and Publications

Certifications

• SystemVerilog Essentials – Udemy (Certification)

• Neural Networks and Deep Learning–Coursera (Certification)

• Supervised Machine Learning: Regression and Classification–Coursera (Certification)

• Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization–Coursera

(Certification)

Publications

• T. Vasei, M. Saber, A. Nahvy, and Z. Navabi, ”An Efficient RTL Design for a Wearable Brain-Computer Interface,” IET Computers & Digital Techniques

• ”Investigating Brain Responses to Transcutaneous Electroacupuncture Stimulation: A Deep Learning Approach” MDPI Algorithms.

• ”A Novel Proof-of-Concept AI-Driven Approach for Advanced Electromagnetic Imaging” Electromag- netics Research

• ”Investigating the Effect of Transcutaneous Electroacupuncture Stimulation on EEG Signals using EEGNet and Saliency Maps” IEEE SPMB (Best Paper Award at IEEE SPMB 2024). Work Experience

Summer 2024 Intern with Digital Medical Experts Inc USA

• Developed a deep learning model for classifying individuals with depression using resting brain signals (EEG), a critical form of time-sequence sensor data, achieving over 90% accuracy.

• This health-related application combined Natural Language Processing (NLP) with Convolu- tional Neural Networks (CNNs) for robust classification. 2020 - 2022 Data Engineer at BRGroup Iran

• Engineered data pipeline optimizations, resulting in a 20% increase in data processing effi- ciency.

• Implemented deep learning models to identify network failures.

• Contributed to the maintenance and troubleshooting of data pipelines, ensuring data integrity and system reliability for deep learning model deployment. Skills

Artificial Intelligence: Machine learning, deep learning, Transformers, Graph Neural Networks (Graph NN), Reinforcement Learning, Generative AI (e.g., VAEs), Multimodal Learning, Self-Supervised Learning, signal processing, biomedical signal processing, brain-computer interfaces. Programming Skills: Python, VHDL, SystemVerilog, R. Verification: SystemVerilog Assertions (SVA), functional simulation, constraint solvers, Universal Veri- fication Methodology (UVM).

Tools: PyTorch, TensorFlow, Keras, Scikit-learn, FPGA design, RTL design. Selected CourseWorks

• Statistical Inference: Applied R programming to analyze real datasets and perform statistical tests.

• VHDL: Modular Design and Synthesis: Designed and synthesized digital cores and systems using VHDL and SystemC.

• High-Performance Computing: Developed parallel algorithms using SIMD, OpenMP, and POSIX.

• Digital Systems Testing and Testable Design: Conducted fault simulation, test generation, and imple- mented BIST techniques.

• Methodologies and Algorithms for ESL Design Automation: Introduction to Catapult High Level Synthesis Tool, Learning BuDDy Software Tool, and Learning Cadane SOC-Encounter Tool.

Contact this candidate