I'm Tanvir Hossain Ovi, an EEG and brain-computer interface researcher. I build deep-learning systems that decode cognition, emotion, and identity from brain signals, and I study how immersion reshapes the way we think.
Research focus
My work sits at the intersection of signal processing and deep learning. I'm mostly asking what EEG can tell us about the mind, and where it quietly stops being able to.
I compare 360° VR and 2D laptop viewing to see how immersion changes mental workload. Theta-alpha ratio and engagement index do most of the talking.
Different EEG frequency bands carry different kinds of information. Some are good at reading what a person is doing, others are better at recognizing who they are. I'm trying to find exactly where that split happens.
EEG-based emotion recognition tends to fall apart when you test it days or weeks later. I build phenotype-aware models that adapt to each person's neural patterns instead of correcting everyone the same way.
Decoding imagined English and Bangla speech from EEG, aimed at assistive communication tools for people with limited motor control.
Selected publications
Machine Learning with Applications · 2026
Pattern Recognition · 2026
Machine Learning with Applications, vol. 22, p. 100752 · 2025
Applied Cognitive Psychology · 2026
From signal to system
Both conditions of the cognitive load study used the same 14-channel dry EEG headset, so the only variable was the display: a 2D laptop screen or a fully immersive VR headset.
360° VR condition
Meta Quest 3 + 14-channel Emotiv EPOC X
2D laptop condition
Laptop display + 14-channel Emotiv EPOC X
Selected builds
Alongside the science, I design and engineer real software end to end: full-stack PWAs I use every day. Proof that I can take an idea from problem to a deployed product.
A shared-mess operations app: daily meals on and off, per-member bazar (grocery) tracking, duty rosters, bills, and an automatic month-end settlement that splits every taka across members.
A privacy-first PWA for organizing sensitive personal data. Every field is AES-256 encrypted and decrypted entirely in the browser with a password that never leaves the device, so the hosted app only ever exposes ciphertext.
A premium multi-page portfolio built on commission for a graphics designer: a bespoke visual identity, an original self-assembling hero animation, and a curated work gallery. Designed, engineered, and shipped end to end.
Reviewing one of my graduate applications, or just curious about phenotype-aware EEG models? I'd love to hear from you.
