Mo Torab Shaikh

Robotics Software Engineer with 6+ years across embedded systems, edge AI, and robotic software. Hands-on with ROS2, C++, Python, PX4, and real-time sensor integration. Building an autonomous drone with visual SLAM, sensor fusion, and state estimation, and deployed multithreaded AI inference on NVIDIA Jetson across 150+ field devices. M.S. in Robotics and AI, University at Buffalo (Jan 2026).

Robotics Software Engineer (Intern) at Nushoor Institute (NGO) (2026-04 – Present)

• Own the full software stack for an autonomous quadcopter (S500 airframe, Pixhawk 6C, M10 GPS, Camera) with NVIDIA Jetson Orin Nano as companion computer.
• Bridged ROS2 Jazzy to PX4 over uXRCE-DDS across uORB topics. Validated flight logic, custom flight modes, and kill-switch protocols in SITL with Gazebo Sim and MAVLink.
• Built a ROS2 package streaming a Sony IMX219 camera into a visual SLAM pipeline (ORB-SLAM3), with checkerboard camera calibration and point cloud visualization. Benchmarked on EuRoC at ATE RMSE 0.036m.
• Integrating monocular visual-inertial odometry (VIO) and EKF IMU/camera sensor fusion for onboard state estimation and GPS-denied navigation.

Graduate Research Assistant at Advanced Materials and Devices Lab, UB (2025-01 – 2026-01)

• Architected a C++ shared-memory pub/sub IPC library for real-time sensor delivery to concurrent ROS2 consumers (Dynamixel finger control, force feedback, visualization) with a ring buffer and zero serialization overhead.
• Resolved torn-read and cross-thread race conditions using a seqlock and atomic operations, ensuring safe concurrent access without blocking producers or consumers.
• Wrote Dynamixel motor control to actuate a 3D-printed finger for repeatable force application against a 20x8 FSR sensor array, with a custom serial framing protocol over UART.
• Applied FFT signal processing and scikit-learn regression to map raw readings to calibrated force values, achieving sub-0.3N mean error on a 20N range.

Tech Lead at Talent Litmus (2021-07 – 2024-07)

• Grew a 4-engineer team to 6 at an early-stage startup, hiring and mentoring 3 engineers to independent productivity; introduced sprint execution and structured code reviews, reducing runtime issues by 90% and technical debt by 70%.
• Owned production infrastructure end-to-end with logging, monitoring, and root-cause analysis, driving production incident resolution and issue mitigation in a fast-moving early-stage environment.
• Scaled platform architecture to 40,000 concurrent users while contributing to 120% client growth over 3 years (58 to 128 clients).

Senior Software Engineer at Helios Web Services: Thinglogix Inc (2018-06 – 2021-07)

• Designed and deployed a Python/C++ multithreaded real-time inference pipeline on NVIDIA Jetson Nano across 150 buses and factories, processing 3 simultaneous RTSP streams per device in fault-isolated threads.
• Used MobileNetV2 with ONNX Runtime for face recognition within strict memory constraints, reducing inference latency by 99%.
• Built offline event buffering for network dropouts, publishing device events to AWS IoT Core via MQTT on reconnect, with automatic reconnection across the bus and factory fleet including Google Coral TPU units.
• Awarded "Excellent Technical Player of the Year" in 2019.

M.S. in Robotics and AI – University at Buffalo (2024-08 – 2026-01)