As a postdoctoral fellow at McGill University, I was involved in a time-varying functional brain connectivity related project using artificial intelligence. More specifically, using multimodal neuroimaging data (simultaneous fMRI-EEG, MEG), I investigated the time-varying characteristics of resting-state networks over different time scales, as well as quantifying the effect of physiological fluctuations (pressure, heart rate, respiration) on fMRI-based resting-state connectivity.

As a postdoctoral fellow at University of Ottawa, I was involved in several projects at Biomedical Signal Processing Lab including: (i) I worked on vital sign monitoring using contactless ultra-wideband radar. More specifically, respiration and heartbeat rates of human subjects were estimated and the posture of targets were detected in a room environment using supervised learning. (ii) In an activity monitoring project, I worked on event recognition problem for home care purposes. More specifically, several algorithms were developed using time-frequency analysiss, time series analysis and deep learning approaches to build an automatic system for detetcion and prediction of fall incidents in senior houses and to assist the elderly living alone.

Organized The 7th IEEE Research Boost: to promote industry driven research.

Organized The 6th IEEE Research Boost: to foster collaborative research projects and networking among the industry, government and academia.

Organized The 5th IEEE Research Boost: to bring together IEEE Members, Academics, Industry, Public Sector and Students to network and enjoy an inspiring technical presentation and discussion.

Course instructor in ENCS6021-Engineering Analysis II.
Delivered lectures to 44 graduate students at Concordia University.

Organized The 4th IEEE Research Boost to foster collaborative research projects and networking among the industry, government and academia, as well as to present some of the top candidates in their field of expertise.

Organized IEEE Montreal Keynote Event: Cars of the Future, to bring together IEEE Members, Academics, Industry, Public Sector and Students to network and enjoy an inspiring technical presentation and discussion.

Organized The 3rd IEEE Research Boost to foster collaborative research projects and networking among the industry, government and academia, as well as to present some of the top candidates in their field of expertise.

As a postdoctoral fellow at Intelligent Signal & Information Processing Lab , I worked on analyzing large-scale data using graph-based signal processing. This work was partnered by Ericsson Montreal with a focus on intrusion detection in cloud systems by online auditing. I also worked on graph-based reachability assessment of the cloud entities in cloud networks. Graph signal processing framework has recently emerged to extend high-dimensional data analysis to networks and irregular domains, and thus, many existing signal/image/video processing, machine learning problems can be revisited from a new graph-based perspective. In view of this, I initiated my research on graph signal processing by investigating the capability of the graph-based filtering for edge-preserving smoothing of images, and accordingly its application in image abstraction and stylization. A unified iterative filtering was developed to perform detail manipulation and fine detail boosting in parallel. In addition, an alternative filtering approach was established in the graph Fourier domain, resulting in noticeable gain in image compression and restoration. Further, I expanded the applicability of the graph signal processing framework by developing an intrusion detection method for distributed sensor networks based on statistical properties of the graph signals. The probabilistic dependence graph was further used for fault detection and localization in smart grid.

As a Lab Instructor at Concordia University, I was responsible for conducting lab tutorials on DC/AC circuits, transformers and DC motor for a group of 20 students. Assisting students in using various hardware tools such as Oscilloscopes, Multimeter and Wattmeter. As a Teacher Assistant and Head TA, I was responsible for training, assisting and supervising a team of 21 teacher assistants, where I demonstrated strong leadership and organizational skills along with a solid work ethics. I coordinated TAs (21 sections), prepared material for tutorial sessions and solutions for assignments, organized midterm exam and prepared the midterm problems, and assigned and coordinated invigilators for the midterm test.

As a research assistant in multimedia signal processing research group at the department of electrical and computer engineering, I conducted research on the processing of image and video signals with special references to their space-time-frequency representations, statistical modeling, developing various estimation and detection techniques, experimentations, and performance studies. Specifically, 1) A unified probabilistic model for the contourlet coefficients of the image signals was developed using the alpha-stable family of distributions. An important motivation for using such distribution was the non-gaussian behavior of the contourlet coefficients of images as well as an appropriate number of parameters that could be used for a better modeling performance. 2) The performance of the probabilistic model was investigated in various types of contourlet representations such as real/complex, and shift-variant/invariant. In particular, the statistics of the data samples were analyzed in terms of the transformation matrices and mathematical operations of the transform. 3) The performance of the developed model was investigated in several estimation techniques such as denoising (reducing additive white Gaussian noise), despeckling (reducing multiplicative and correlated noise), deblurring (restoration from the degradation due to point spread function), and fusion (obtain an informative signal from multi-sensed data). Estimation performances were evaluated on several data sources that include natural images and video, medical images, and remote sensing images. 4) The performance of the developed model was also studied in the case of the detection of edge and textures and the detection of watermark for copyright protection.