Hussein Sarwat Email & Phone Number

Shanghai, China

Shanghai, China

· Optimized data acquisition from internal databases by developing a RAG model that extracts queried information.· Enhanced user experience and information retrieval efficiency by developing a browser extension that utilizes LLMs to summarize web content.

Shanghai, China

· Achieved 84.8% accuracy in finger gesture classification and 85.4% in grip force estimation by designing anddeveloping a wristband using Hall effect sensors, enhancing precision for real-time interactions· Improved model accuracy over single-method approaches by developing and testing a multimodal architecture in PyTorch, integrating both raw data and extracted features to optimize real-time performance· Increased dataset diversity and analyzed inter-subject performance using GAN and data augmentation techniques· Implemented few-shot transfer learning using prototypical networks in PyTorch, allowing models trained onprevious subjects to adapt effectively to new subjects with comparable performance, minimizing retraining needs· Optimized data collection and device synchronization for real-time signal processing by integrating multiplesensors on an STM32 microcontroller using I2C communication in Embedded C· Refined input data for machine learning performance by conducting multi-domain feature extraction (time, frequency, time-frequency) from FMG, EMG, and IMU signals, resulting in high-quality, versatile data processing· Boosted gesture recognition reliability in real-time environments by creating an ensemble classifier with post-processing techniques, significantly enhancing model consistency and accuracy· Developed a real-time GUI interface in PyQT5 to display current gestures, sensor readings, and dynamic signal plots, enhancing usability and providing immediate, visual feedback for gesture and sensor data· Developed an experimental protocol to collect data from 21 subjects for model training and implemented a real-time testing configuration, ensuring robust model validation and improved accuracy under real-world conditions

Cairo, Egypt

· Enabled adaptive ’pick-and-place’ tasks on various objects by training a 6-DOF robot using TensorFlow for object detection and classification, improving robotic handling in dynamic environments· Optimized robotic trajectory planning by utilizing the ROS MoveIt package, enhancing task efficiency and accuracy in complex movement scenarios· Developed and improved undergraduate course content on mechatronic systems, robotics, machine learning, mathematical modeling, and control, actively seeking innovative teaching approaches, which led to 84% positive student feedback· Fostered interdisciplinary collaboration within the Engineering department by delivering seminars on machine learning, machine vision, and robotics to faculty members, supporting knowledge expansion across disciplines

Cairo, Egypt

· Used self-organizing maps to cluster individuals by health metrics, with healthy individuals centrally positioned and unhealthy individuals in outer clusters, improving pattern identification in unsupervised health assessments· Enhanced classification accuracy for health assessment by applying XGBoost for supervised learning on sensor data, supporting effective differentiation of health states· Published two conference proceedings on applications of machine learning applications and a journal paper on machine learning, embedded systems, and IoT applications in medical diagnosis and assessment· Collaborated with hospitals and medical professionals to collect data from post-stroke patients, enhancing the dataset’s clinical relevance and supporting model development for health monitoring applications· Built a mobile app for real-time storage and display of gesture data, enhancing accessibility and usability ofgesture recognition results for end-usersAchieved 85% gesture recognition accuracy by developing a data glove with 16 integrated sensors, enabling precise tracking for health monitoring applications

Ontario, Canada

· Reduced navigation time of autonomous mobile manipulators by up to 67% and achieved an 80% success rate inspecific scenarios by optimizing algorithms and enhancing navigation efficiency· Developed and implemented adaptive control algorithms using ROS packages to navigate a mobile manipulatoraround movable obstacles, leveraging local cost-map data for real-time adaptability· Created a navigational behavior in CMake and Python, integrating C++ libraries to streamline interactions and ensure robust functionality in complex robotic systems· Simulated and tested navigation algorithms in Gazebo, validating their effectiveness in real-world applications and optimizing manipulator performance under controlled conditions.· Coordinated software updates using a Git repository, facilitating collaboration among a team of over six members on mobile manipulator development and version control· Engineered an algorithm with an average accuracy of 0.1m to calculate the center of square obstacles, improving trajectory planning and obstacle avoidance for safer navigation

Cairo Governorate, Egypt

Cairo Governorate, Egypt

Master'S Degree, Mechanical Engineering, 3.07

Activities and Societies: Multimedia Team for International Student Service Center

Bachelor'S Degree, Mechatronics, Robotics, And Automation Engineering, 3.15

Activities and Societies: Magazine Writer/Editor at Astronomy Club