Muhammad Ahmed

Robotics, Embedded Systems, and Hardware Integration Engineer

Robots should work outside the demo.

I connect sensing, embedded software, motion planning, and physical hardware to build robotics systems that can be tested honestly.

View selected work Resume View GitHub Connect on LinkedIn

About

I am a robotics and mechanical engineer with an M.S. from Duke University. I enjoy the hands-on work of bringing a system together: designing a mechanism, connecting its electronics, making sense of its sensors, and getting the software to move real hardware.

My projects span robotic manipulation, embedded sensing, assistive devices, and human-robot interaction. I document what I built, what I tested, and what still needs work so others can understand and build on it.

Education

Duke University · May 2026

M.S. Mechanical Engineering & Materials Science

GPA 3.74 / 4.0

Robotics, manipulation, medical robotics, scientific computing, and experimental design.

Duke University · Graduate Certificate

Medical robotics

Medical and surgical robotics, robotic manipulation, and simulation.

SRM Institute of Science and Technology · May 2023

B.Tech. Mechanical Engineering

CGPA 8.18 / 10

Mechanical engineering foundation with research in tribology and materials characterization.

Research & publication

Tribology & materials

Research Assistant
SRM Institute of Science and Technology
April 2022 to September 2023

Tribo-Corrosion and Mechanical Performance of Electro-deposited Nano-Composite h-BN/Epoxy Coating

Proceedings of the Institution of Mechanical Engineers, Part J: Journal of Engineering Tribology · 2024

Studied coatings for wear and corrosion resistance through controlled lab experiments. My work included pin-on-disc testing, electrochemical impedance spectroscopy, FTIR, experimental design, and statistical analysis.

Read the paper

Selected work

Kinova Gen3 Lite arm after releasing a ping-pong ball toward a cup target.

Kinova Gen3 Lite arm in its wind-up pose before throwing a ping-pong ball.

Gazebo view of the Kinova Gen3 Lite pick-and-place stacking scene.

RViz view of the Kinova Gen3 Lite pick-and-place stack and planning scene.

Outcome: Built a ROS 2 manipulation suite spanning a physical throwing demo plus related pick-and-place and writing prototypes in simulation.
What I built: Developed the ROS 2 workflows for throwing, block stacking, and writing initials, from task logic and MoveIt 2 planning to gripper sequencing, collision scenes, joint-limit configuration, and RViz visualization.
Proof: Physical demo video plus Gazebo and RViz captures and repositories. The throwing path uses joint-state-triggered gripper release and open-loop targeting; no repeatability or hit-rate is claimed.

Technologies

ROS 2
MoveIt 2
C++
Kortex
RViz
Gazebo

Beer Pong Demo video Pick and Place Write Initials

White 3D-printed NeuroBot prototype with a front-facing camera on a sunlit table.

Outcome: Implemented a Jetson-oriented perception pipeline that adapts face, gesture, and audio outputs for late fusion.
What I built: Built the companion software and adaptive-sensing pipeline, including face processing, voice interaction, gesture and audio adapters, fusion logic, session logging, and tools for model evaluation, fault injection, and Jetson latency/power profiling.
Proof: Repository and compute-cost artifact across four Jetson power modes; model accuracy, clinical inference, ROS 2 actuation, and validated multimodal improvement remain unproven.

Voice sessions include documented cloud-service boundaries.

Technologies

Python
PyTorch
OpenCV
MediaPipe
NVIDIA Jetson

View repository

Photograph of the CRUTCH prototype assembly used for exploratory bench work.

Outcome: Developed separate bench prototypes for sensing, logging, stepper actuation, and a compliant spring-damper concept.
What I built: Designed the spring-damper concept and 3D-printed mounts, and developed separate bench prototypes for ESP32 telemetry, load and motion sensing, stepper control, and Python logging and visualization.
Proof: Repository and prototype photo. The committed paths are separate prototypes, not a verified integrated system, force-reduction result, clinical device, or safe human-use claim.

Technologies

ESP32
Embedded C++
IMU
Load sensing
Python logging
Mechatronics

View repository

CAD perspective view of the six-joint ArmBot concept.

Outcome: Created a six-joint robot-description concept for inspecting a custom arm in ROS 2.
What I built: Designed the six-joint arm concept in CAD and built its Xacro/URDF description, including seven link meshes, joint definitions, collision and inertial properties, and RViz/Gazebo launch configurations.
Proof: The repository documents the robot description and a current clean-launch blocker; no demonstrated control system is claimed.

Technologies

URDF
Xacro
ROS 2
CAD
RViz

View repository

Experience

Robotics Systems Engineer

24 and Up

May 2026 to present

Work across embedded software, sensor integration, hardware-software interfaces, rapid prototyping, system testing, and real-world debugging for robotics and intelligent automation systems.

Junior R&D Engineer

Duke University Wilkinson Garage Lab

June 2025 to May 2026

Contributed to robotics and medical-robotics prototyping, assistive-device concepts, sensor integration, and control work; supported fabrication and troubleshooting with 3D printers, CNC machines, laser cutters, and electronics workbenches; trained students in safe lab-equipment use.

Teaching appointments

Duke University

2024 to 2026

Supported Robotics & Automation, Experimental Design & Research Methods, and Thermodynamics through lab and office hours, ROS assignment troubleshooting, design and test planning, technical feedback, and engineering problem-solving.