Hello! I am Tahsin. I completed my Masters degree in the ECE department of University of Florida.
I graduated in Electrical and Electronic Engineering from Bangladesh University of Engineering and Technology (BUET) in 2017. After graduation, I worked as Machine Learning Engineer in Semion Limited and AI Samurai, served as an RA in the MHealth Lab, Department of Biomedical Engineering, BUET and mentored several undergraduate students in their thesis and research works.
I am currently looking for job opportunities in the domain of AI.
Developed an algorithm for small object detection in CT images using Dense Atrous Spatial Pyramid Pooling and a Spatial Context Network with Reverse Axial Attention.
Co-developed an acuity assessment pipeline for ICU patients incorporating explainable AI algorithms to identify features contributing to the worsening of patient conditions.
Developed a semi-supervised object detection pipeline for electric pole detection from car dashboard camera images using Fast-RCNN with ResNet50 backbone and YoloV4.
Co-developed a semi-supervised CNN model for abnormality detection in dairy product images.
Co-developed SemRad, an inference tool and Class Activation Mapping (CAM) tool for localization of abnormalities in chest X-ray images using ResNet101.
Developed an Amazon Alexa skill to help users detect diseases from symptoms.
Worked on developing the backend deep learning model for RadAssist, assisting doctors in detecting and localizing abnormalities from brain CT scan images.
A complete teleradiology solution that incorporates deep learning models to detect chest abnormalities. The framework is JavaFX and SQL is at the backend. This was a professional project. I designed an inference tool for abnormality detection and localization in Chest X-Rays based on MobileNet Algorithm. This was later integrated into the software.
A complete teleradiology solution that incorporates deep learning models to detect chest abnormalities. The framework is JavaFX and SQL is at the backend. This was a professional project. I designed a Class Activation Mapping tool based on this paper for localization of abnormalities in Chest X-Rays. This was later integrated into the software.
This project was performed on the DRIVE dataset, using a U-Net-based architecture with residual blocks. It was part of a research work which later got accepted at the 4th IEEE WIECON-ECE Conference 2018. Details are available in the Publications section.
Using the Named Entity Recognition (NER) method described here, I developed a model that can extract pathology terms from EHR reports. This is an ongoing research project.
I developed a device using Raspberry Pi, Arduino, and a transformer that can measure the total harmonic distortion (THD) present in the power system. This was a part of our undergraduate thesis.
This app was designed to help children stimulate their cognitive development. It uses an Inception v3 model as its backbone, pretrained on the ImageNet dataset and fine-tuned on additional data. It was part of a research work which later got accepted at the 2017 IEEE Region 10 Humanitarian Technology Conference (R10-HTC). Details are available in the Publications section.
This Alexa skill can recite all differential diagnoses for user-fed symptoms. It includes a comprehensive list of 1700 differential diagnoses and their corresponding symptom mapping. The development of this skill is still in progress.
Using bidirectional LSTM and CRF-based entity recognition model, risk factors were identified for heart diseases in diabetic patients. The dataset was from i2b2. This was a professional project.
This Android app helps easily navigate the vast landscape of differential diagnoses, aids medical students in learning, and helps normalize and standardize communication between physicians. The app is available at Google Play.
A 2-D Car racing game for PC and Android devices (Phone, Tablet, Smart TV etc.) with an external control system and a user-friendly gaming interface was designed. Detailed explanation and demo are available here on Youtube. This was an academic project for the Control System Lab.
Arduino along with a GSM module, LCD panel, proximity and sonar sensors were used for this project. This was an academic project for the Communication I Lab.
This was the final project for the Measurement and Instrumentation Laboratory. I made a 2-channel oscilloscope that can simultaneously measure and plot voltage and current values on a Graphical User Interface designed by me. It was also the last undergrad project I worked on.
The infamous Flappy Bird game could be played by waving a hand in this project. Arduino along with a SONAR sensor were used for this project. A demo is available here on Youtube.
This was a group project. We made a home automation system that had electric fans and lights as test loads, and an Android app was built to control the loads from a distance. This prototype used Bluetooth to connect.
Mentored an undergraduate student in his thesis work titled "COVID Infection Analysis via Lung Lobe Segmentation using Deep Learning".
Supervised two high-school students to get them familiar with research work in hardware security and machine learning under the Student Science Training Program (SSTP) at University of Florida..