Implementing and improving existing biomechanical algorithms for gait analysis and phase/subphase detection at the firmware level.Exploiting multiple sensors and sensor fusion techniques for enhanced gait analysis.Developing real-time gait detection algorithms and production-ready codes, adhering to production constraints.Contributing to all phases of digital system design, including functional specification, firmware development, and troubleshooting.

Process control, automation and data analysis using deep learning and reinforcement learning for optimizing water treatment process of the town of Drayton Valley. Developed a water treatment simulator using advanced machine learning models on extracted datasets.Achieved 40% enhancement in water production efficiency through optimized decision-making with a reinforcement learning agent. Oversaw model deployment into the production system; designed unit tests and prepared the CI/CD pipeline.Served as the main point of contact in weekly client meetings, providing project updates and receiving valuable feedback from domain experts. Proficiently communicate complex technical concepts to non-technical stakeholders, ensuring clarity and understanding through data visualization techniques and dashboards

Machine-Learned Adaptive and Predictive Real-time Controller Design for Lower-Limb Exoskeletons (Thesis Project):- Designed an adaptive controller using reinforcement learning techniques, such as temporal difference (TD) learning, and general value functions (GVFs) for multimodal prediction of auxiliary tasks, specifically tailored for lower-limb exoskeletons. The controller accurately predicted users' intentions during transitions between multiple operating modes. The communication between devices was established seamlessly using CAN bus and UDP/IP protocols.- Achieved an average 50% significant reduction in the total number of required user switching actions compared to the best possible non-adaptive approach. This improvement enhances users' trust in the system and optimizes human-robot interactions, reducing the effort required.

Developed a biologically-inspired state controller for coordinating the activation of electrodes placed in the spinal cord of animals using Intraspinal Microstimulation (ISMS) to restore walking ability after complete spinal cord injury.Leveraged feedback from ground reaction forces and angular velocities of hind limbs to accurately differentiate walking sub-states.This project represents a significant milestone in the clinical translation of ISMS. If successful, it has the potential to profoundly impact the lives of thousands of individuals living with spinal cord injuries, granting them the ability to regain independent walking capabilities.

Developed advanced machine learning techniques to address the control and acceptance challenges of myoelectric prostheses.Created a position-aware controller utilizing a convolutional neural network (CNN) and transfer learning.The implemented strategy significantly reduced the re-training requirements for individual users, while maintaining a high level of accuracy in the trained neural network model (92% classification accuracy).

Tutoring and supervision of laboratory activities and students' coursework.