I'm Aditya Mohan - a Computer Science graduate and a Software Engineer. I'm currently enrolled in the Master's of Science in Computer Science from Arizona State University, and parallelly, working as a part-time Research Assistant. Most recently, I worked with Amazon Web Services in the AWS Proactive Security organization to build a highly secure, optimized and scalable tool to facilitate on-demand network scans throughout the extensive AWS internal network. Prior to that, I worked at Oracle Corporation as a Software Engineer in the Oracle Analytics division. I'm interested in large-scale cloud platforms and machine learning and I love to work in areas pertaining to cloud infrastructure and machine learning. I firmly believe in the importance of such scalable systems given the "data explosion" and the need to securely maintain and make sense of valuable information from disorganized data. Apart from my research interests, I enjoy taking up the occasional side projects and venturing into new realms of Computer Science. Outside of work, I enjoy spending time playing instruments (particularly the guitar), following/playing football, travelling and dabbling in some trivia.
Automating the analysis of volcanic features in time-series ASTER satellite imagery to identify subtle thermal anomalies for over 7k volcanic time-series data points using Local RX and PyTorch-based data parallel CNN models.
Worked on reconstructing multi-view HDR images from single exposure LDR images using a Deep Learning algorithm while achieving consistency among multiple views. A poster for this paper has been accepted to Eurographics '22!
Built a performance prediction Java based tool using a combination of static and dynamic analysis of the CUDA Program and obtained features for performance prediction using the NVIDIA Visual Profiler. Improved the GPU execution time prediction by building memory access penalty models using a Java application.
A PDF Renderer desktop application with special functionalities for a client. The renderer includes an annotation side-bar which displays the definition of a word by looking up previous occurrences of the same word through local database transactions.
This project was carried out as part of the "Pixxel Open Data challenge". The task was to segment hyperspectral satellite imagery using an unsupervised classification technique. The 242 bands of satellite imagery data was collected by NASA's EO1 program.
An online repository for various image processing and manipulation techniques applied on some photos captured by myself. Looking forward to expanding the repository to incorporate more techniques in Computational Photography.
This project was part of my internship. Apart from a ton of literature survey on employing ML techniques, I came up with a simple ML classifier which makes use of the Random Forest Classifier to classify leather samples into defective and non-defective samples.