roysrc Goto Github PK

I am interested in building tools that allow us to measure the safety, trustworthiness, and fairness of AI systems for safety-critical, regulated, and scientific domains such as in biomedical and healthcare applications. I am also interested in creating AI systems that intrinsically satisfy the above conditions of safety, trustworthiness, and fairness when deployed to real-world applications. I am particularly interested in exploring ideas from programming languages and software engineering, such as neuro-symbolic programming, program verification, and program synthesis, to achieve this goal.

I am also interested in the application and development of machine learning algorithms for algorithmic and structured data problems such as graphs, sequences, images, and time-series data. Of particular interest are applications of these ideas to solving problems in AI for science, computational drug design, healthcare, and biomedical informatics.

I lead the Quantitative Light Imaging Laboratory's efforts in software development for data acquisition and analysis, including software-hardware integration and development and deployment of AI tools for solving biomedical problems.

My project revolved around developing the next generation of AI embedded intelligent microscopy systems for non-destructive sample imaging through the integration of ML with SLIM (Spatial Light Interference Microscopy) imaging modality.

I was primarily involved in implementing new and optimizing existing group theory algorithms for the Orbiter computer algebra system (An algebra system for the classification of combinatorial objects) for both CPU and GPU to be run on Summit High Performance Compute Clusters. I optimized one of the core graph theory algorithms used throughout the software for performance improvement of an order of magnitude by using efficient data structures. I have also used machine learning techniques such as deep reinforcement learning to optimize tree-based data structures used to represent the orbits of groups acting on sets in real-time. My approach achieved state-of-the-art results compared to existing algorithms, primarily the Seress shallow Schreier tree algorithm in terms of memory usage and the quality of the generated tree. I have presented this research at the 50th Southeastern International Conference on Combinatorics, Graph Theory & Computing.

I assisted in teaching the Operating Systems course at Colorado State University. I helped write class assignments, held help sessions and office hours and assisted with class logistics. The assignments ranged from writing scheduling algorithms to multi-process and multi-threaded programs. I was also involved in building an automated grading system to grade student submissions using error carried forward logic.

NOTE: Top languages do not indicate my skill level or anything similar to that, it's a github metric of which languages I have the most code on github.