Hi, I'm Bassim Beshry

About Me

My name is Bassim Beshry and I’m a Computer Science graduate from UBC with a strong passion for networking and infrastructure engineering. I enjoy working close to the systems configuring and troubleshooting networks that are reliable, secure, and scalable. I value clean configurations, strong fundamentals, and automation where it meaningfully improves reliability and operational clarity.

My interest in networking took shape during my time in the Aerial Robotics club, where I spent two years working on distributed and autonomous systems. Through this experience, I gained exposure to computer vision pipelines, inter-process communication, and networked systems using tools like Docker and ZeroMQ. These projects emphasized the importance of low-latency communication, fault tolerance, and observability in real-world systems.

After graduating, I expanded this interest through homelabbing, where I actively design and manage networked environments involving routing, monitoring, and containerized services. Exploring cloud platforms further strengthened my understanding of hybrid networking concepts such as VPCs, serverless architecture, and infrastructure segmentation. I’m motivated by roles that allow me to build and maintain resilient network infrastructure, support on-prem and cloud integrations, and contribute to systems that improve performance, security, and operational efficiency.

Outside of work, I enjoy lane swimming, sim racing, and videogames with expansive worlds and complex narratives. My favorite book is The Sound of Waves by Yukio Mishima, and my favorite games include Destiny 2, the Nier series, Final Fantasy 14 and Clair Obscur: Expedition 33.

AI Developer @ FlyWheel Solutions

• - Developed a two-pass LLM classification system using the OpenRouter API to extract structured attributes from industrial equipment descriptions and test model efficiency, enabling automated categorization across 4 hierarchical levels and 25+ attributes
• - Built an automated LLM prompt benchmarking and evaluation pipeline in Python, tracking PDF parsing accuracy across datasets to standardize model performance measurement and identify cost optimal models
• - Iteratively refined prompts using Llama Cloud Services, improving table extraction robustness and increasing parsing accuracy by 60%, significantly enhancing downstream data quality

Software Developer @ UBCO Aerial Robotics and Rocketry Club

• - Developed an autonomous UAV flight procedure using MavSDK, C++, and Mission Planner simulations to retrieve water from a tank and navigate waypoints for wildfire suppression
• - Designed and implemented an autonomous spiral search algorithm for aerial fire detection during ground surveys
• - Troubleshooted cellular communications between the ground station and drone in order to stream FPV video footage
• - Worked with ZeroMQ for inter-process communication between the drone and ground control station.
• - Developed a camera calibration algorithm with OpenCV to fix radial distortion seen in webcam lenses
• - Trained a YOLOv7 object detection model using an online dataset and OpenCV to identify UAV landing pads; conducted supporting research on LiDAR technology for enhanced precision during autonomous landing