.jpg)
Education
- Ph.D in Electrical Engineering (Concentration: Computer Engineering), The University of Texas at San Antonio, August 2022
- M.S. in Electrical Engineering (Concentration: Computer Engineering), The University of Texas at San Antonio, December 2021
- B.Sc. in Electronics and Communication Engineering, Khulna University of Engineering and Technology, September 2014
Research Experience
Graduate Research Assistant
Employer: IoT Security Lab, The University of Texas at San Antonio
Supervisor: Dr. Guenevere (Qian) Chen, Assistant Professor, Electrical and Computer Engineering
Cyber Physical System
- Conducted driving tests in simulation-based Testbed using OpenDS for 50 college-age drivers (IRB approved) under 20 driving tasks and 12 cyber-attack scenarios.
- Collected physical and behavioral data of the vehicle (e.g., position, steering angle, reaction time etc.).
- Proposed Cyber, Physical and Human factor-based framework, ExHPD for driving behavior modeling to detect vehicle cyber-attack using Random Forest and LSTM Autoencoder model.
Enterprise Network
- Developed testbed of bare-metal servers for host log (benign/malware) data collection with FOG-project, WLS, Windows ETW and ELK stack.
- Collected audit and application log (Windows/Linux) dataset under benign scenario for 90-days and 35-users (IRB approved) in a large enterprise network in collaboration with Sandia National Lab (SNL).
- Collected malware dataset in controlled environment (Cuckoo sandbox) for more than 150 malware samples (e.g., Adware, Ransomware, Backdoor/Trojan etc.).
- Indexed collected data (2TB) in Elastic server using Logstash, analyzed and visualized using Kibana for preprocessing (anonymization for privacy) in Python.
- Designed and implemented DeepRan an attention-based bi-LSTM and CRF model for ransomware early detection and classification with more than 98% accuracy.
- Proposed LogGNN a Graph Neural Network (GNN) based graph embedding algorithm for representation learning of heterogeneous Provenance graph constructed from host log and behavioral data.
- Developed Cyber-Psychology (Delay Discounting, Risk-Taking) mapping framework for early detection of Insider Threat.
- Currently working on GNN-LSTM based unsupervised malware detection model using provenance graph constructed from collected malware logs for threat hunting in enterprise network.
Graduate Research Assistant
Employer: The University of Texas at San Antonio, United States
- Designed differential privacy mechanism for publishing optimized building energy consumption data.
- Analyzed k-anonymity, Local differential privacy (LDP), Exponential and Laplace mechanism for differential privacy mechanism and.
- Analyzed differential privacy mechanisms for social graphs using Facebook data from SNAP
Undergraduate Research
Employer: Khulna University of Engineering and Technology
- Studied and surveyed ultrasound imaging systems in biomedical imaging
- Developed ultrasound strain imaging system using dynamic programming and information theory for nonlinear ultrasound RF data and published two conference papers
Skills and Expertise
- AI Algorithms: ML (SVM, LR, RF), DL (CNN, LSTM, GNN, Autoencoder, GAN), NLP, Transformer
- Programming Languages: Python, C, C++, MATLAB, CUDA
- Software & Tools: ELK stack (Elasticsearch, Logstash, Kibana), WLS, WireShark, TensorFlow, PyTorch, Keras, Weka
- Web & Cloud: Google Colaboratory, Azure Notebooks, IBM Watson Studio
- Platforms: Windows, Linux/Unix, MacOS
Projects Accomplished
Graduate Projects
Institution: The University of Texas at San Antonio
- Driver Behavior Analysis Model: Developed driver behavior modeling tool using deep learning autoencoder models with real life driving and driver behavioral data.
- Smart and Secured Parking System: Developed RF based smart parking system with light weight MQTT protocol and performed security vulnerability analysis with Wireshark in IoT Security course.
- TRN for Video Summarizing: Implemented multiscale temporal relational reasoning network (TRN) for video event detection and summarizing in Deep Learning course.
- Cache Performance Simulator in Python: Designed Cache Performance Simulator in Advanced Computer Architecture course in python
- Cloud Solution for Medical Emergency: Proposed and implemented a cloud solution for handling medical emergency visits in rural areas through android application. Implemented collective communication system in cloud using OpenStack and interfaced with android application in Cloud Computing course.
Undergraduate Projects
Institution: Khulna University of Engineering and Technology
- Designed and implemented land rover robot controlled by mobile commands using DTMF
- Designed and implemented a microcontroller-based PC remote controller system with RC5 protocol
- Designed and implemented Line follower robot with mash solving ability
- Developed FPGA based 64-bit magnitude comparator with BIST facility
Publications