Experience

Five years of measuring minds

From robotics competitions to EEG laboratories, here is the education, research roles, and skills behind the work.

Education

Bachelor of Science, Electrical & Electronic Engineering

University of Chittagong, Bangladesh

Jan 2020 to Nov 2025

BSc in Electrical and Electronic Engineering

CGPA: 3.78 / 4.00
Merit position: 2nd out of 63 students

Graduation thesis

EEG-based cognitive load analysis in 360° virtual reality and 2D laptop video modes

Research appointments

Where the work is done

Research Assistant, Biometric Intelligence and Vision Laboratory

Jul 2025 to Present

Department of CybersecurityState University of New York, Canton, USARemote

Preprocessing EEG recordings, designing neural network architectures, feature selection, statistical validation, and drafting manuscripts.
Supervised by Dr. A M Mahmud Chowdhury, with Dr. Nazmul Siddique of Ulster University, UK, co-supervising the neural signal work.
First author on the four studies produced during the appointment.

Undergraduate Researcher, University of Chittagong

2023 to 2025

Department of Electrical and Electronic EngineeringChattogram, BangladeshIn person

Recorded EEG from participants for four studies: the VR thesis, imagined Bangla words, emotionally valenced Bangla words, and audio vs. video calls.
BSc thesis: designed a VR vs. laptop video learning protocol and recorded 30 participants on a 14-channel Emotiv EPOC X. VR lowered cognitive load by 21% and raised engagement by 59%.

Research experience

Four studies on what EEG really carries across sessions, channels, and people

Complementary Raw and Engineered Representations for Focal and Non-Focal iEEG Classification

2026

Designed a recording-level evaluation protocol that prevents windows from the same Bern-Barcelona recording from appearing across training and test splits.
Built CSRA-EF Net by fusing a shared cross-channel raw-iEEG encoder with 280 engineered differential-entropy, spectral-entropy, Hjorth, and inter-channel-correlation features.
Reached 96.27% accuracy and 0.994 AUC across three training seeds; ablation showed the engineered-feature branch was the only component with a detectable effect under a pre-declared one-point equivalence margin.

Phenotype-Aware Personalization for Cross-Session EEG Emotion Recognition

2025 to 2026

Investigated cross-session performance degradation in EEG-based emotion recognition by analyzing the SEED-IV dataset across three weekly recording sessions per participant.
Implemented variance decomposition and two-way ANOVA, which placed 80.1% of the explained variance in the participant-by-emotion interaction, so cross-session variability reflects structured individual differences rather than random noise.
Conducted brain-behavior correlation analysis and hierarchical clustering to identify distinct user phenotypes affecting cross-session model generalization.
Developed phenotype-aware calibration that raised cross-session accuracy from 39.96% to 88.71%, outperforming uniform correction methods.

EEG-Based Frequency Band Analysis for Cognitive State and Biometric Applications

2025

Independently analyzed a publicly available EEG dataset of 20 participants across three weekly recording sessions to investigate frequency band dissociation between task classification and person authentication.
Implemented a complete preprocessing pipeline with ICA artifact removal and computed cognitive engagement metrics including Theta-to-Alpha Ratio and Engagement Index.
Developed and evaluated EEGNet with Squeeze-and-Excitation blocks, Channel Mixing variants, and EEGPT architectures for cognitive state classification across seven frequency bands.
Conducted cross-session subject authentication experiments, discovering that delta oscillations achieved 86.9% task classification accuracy but only 15.6 to 23.6% authentication accuracy, while broadband signals reached 69.9% authentication performance.

Cognitive Load Analysis in Virtual Reality and Traditional Display Environments

2024 to 2025

Designed an experimental protocol comparing cognitive load during 360° VR immersive viewing versus 2D laptop display with 30 participants using a 14-channel Emotiv EPOC X.
Developed a preprocessing pipeline incorporating bandpass filtering, ICA artifact rejection, and signal normalization, then extracted Theta-Alpha Ratio, Engagement Index, and Beta-Alpha Ratio.
Performed statistical validation using MANOVA and ANOVA with FDR correction and built classification models with SVM, Random Forest, and kNN using five-fold cross validation.
Demonstrated that VR environments reduced cognitive load by 21% while increasing engagement by 59%, achieving 83.5% classification accuracy with TAR identified as the optimal marker.

Toolkit

Technical skills

Programming & Frameworks

PythonTensorFlowKerasPyTorchScikit-learn

Signal Processing

Emotiv EPOC X recordingMNE-PythonEEGLABICA artifact removalEEG preprocessingFeature extraction

Statistical Analysis

t-testWilcoxonANOVAMANOVAFDR correctionHierarchical clusteringData modeling

Computer Vision

MediaPipeOpenCVROI extractionNFIQ2

Data Visualization

MatplotlibSeabornTableau

Tools & Platforms

LaTeXGoogle ColabGitLinux

Simulation

Advanced Design System (ADS)RSoftPSpice

Recognition

Honors & awards

First Runner-up, Student Merit Award
EEE Fest, University of Chittagong
Feb 2025
Champion, Robo Soccer Competition
EEE Fest, University of Chittagong
Jan 2023
Second Runner-up, Robo Soccer Competition
Engineering Day, University of Chittagong
Jun 2022
Principal's Award for Academic Excellence (4th position, Science group)
Dhaka Imperial College
Sep 2018
Government Junior Scholarship
Junior School Certificate examination, Chatkhil Panch Gaon Government High School
2014

Language

Standardised tests

IELTS Academic

14 April 2026

7.5overall band

Listening: 8.0
Reading: 8.0
Writing: 6.0
Speaking: 7.0

Want the full picture?

The complete CV has publication-by-publication detail, references, and contact information.

Get in touch Download CV