Muhammad Wasiq Saleem

Muhammad Wasiq Saleem

Robotics & AI Engineer | Computer Vision & Embedded Systems Specialist

Based in: Islamabad, PakistanMost recently: President, Team ALIF (Autonomous Racing), NUSTExperience: 2 years

Hire me

Mechatronics Engineer specializing in autonomous systems and embodied AI. I have a proven track record of developing high-performance robotics pipelines, including achieving a 5x increase in inference speed for depth estimation and reducing hardware costs by 80% for autonomous racing prototypes. Experienced in leading multidisciplinary teams and designing robust, sensor-efficient systems for hazardous environments.

Experience

President

Team ALIF (Autonomous Racing), NUSTIslamabadworkSep 2025
reduction in hardware costs
80%reduction in hardware costs

Summary

Lead a 10-member multidisciplinary student team designing and building an autonomous race vehicle. Mentor and coach team members across mechanical design, electronics, coding, and robotics integration, guiding them from CAD through working hardware via hands-on sessions and project reviews. Built a functioning, student-led robotics team culture - coaching peers hands-on in robotics, coding, and AI fundamentals.

Results

  • Built a functioning autonomous racing prototype using only an ESP32 and a smartphone camera.
  • Eliminated the need for expensive sensors (LiDAR, depth cameras, Raspberry Pi), reducing hardware costs by ~80%.
ESP32Smartphone Camera IntegrationCost Optimization

Embodied AI & Computer Vision Research Intern

NCAIIslamabadworkFeb 2025 – Aug 2025
FPS real-time processing speed
5FPS real-time processing speed
increase in depth estimation accuracy
50%increase in depth estimation accuracy
increase in inference speed (FPS)
5xincrease in inference speed (FPS)

Summary

Developed object detection, depth-perception, and semantic mapping pipelines for robot navigation. Integrated YOLO and monocular depth estimation into a real-time robotics perception pipeline; managed datasets, evaluation, and model optimization. Delivered a real-time, low-cost perception system that is a kind of hands-on build ideal for teaching AI and robotics concepts to students.

Results

  • Achieved a processing speed of 5 FPS for the integrated YOLO and monocular depth estimation pipeline.
  • Fine-tuned Depth Anything V2 for RGB-based mapping, increasing depth estimation accuracy by ~50% and achieving a 5x increase in inference speed.
YOLOComputer VisionMonocular Depth EstimationSemantic MappingRobot NavigationDepth Anything V2Model Fine-tuning

Hardware & Robotics Systems Intern

NCAIIslamabadinternshipSep 2024 – 2025
Built ROS 2 control plugins and closed -loop motor control for ESP32-based differential-drive robots. Integrated LiDAR and wheel encoders for autonomous navigation; developed a serial/WiFi hardware interface bridging embedded firmware and ROS 2. Delivered a low-cost, reliable robotics platform, cutting hardware cost by roughly 80% while keeping navigation performance intact - the kind of build-it-yourself project ideal for teaching.
ROS 2ESP32LiDARWheel EncodersWiFiSerial InterfaceEmbedded Firmware

Skills

Technical

PythonPython
YOLO
ROS 2
ArduinoArduino
C++C++
ESP32
CAD
OpenCVOpenCV
SolidWorks
PyTorchPyTorch
CNNs
Vision Transformers

Areas of expertise

Project-Based Learning
Technical Mentoring
STEM Instruction
Public Speaking

Projects

C++ Programming Chatbot

Developerpersonal
Built a chatbot in C++ to answer and explain programming concepts, aimed at improving beginner learning outcomes.
C++

Computer Vision & Embodied AI Research (NCAI)

Researcherpersonal
Built object detection and depth-perception pipelines for robot navigation using YOLO and monocular depth estimation.
YOLOMonocular Depth EstimationComputer VisionDepth Anything V2

ROS 2 Differential-Drive Robot

Lead Designeracademic
Designed a low-cost ROS 2 differential-drive robot from CAD to deployment, creating a complete educational platform for teaching embedded systems, sensors, control, and navigation.
ROS 2CADEmbedded SystemsSensorsControlNavigation

Smart Motorbike Tracking System

Developerpersonal
Built a real-time, multi-threaded Python system using YOLO object detection, speed tracking, and a GUI - an engaging, visual demo of AI concepts for students.
PythonYOLOGUIMulti-threading

Arduino-Based Obstacle Avoidance Robot

Developerpersonal
Developed an autonomous differential-drive robot using Arduino and ultrasonic sensors, a classic beginner-friendly build for teaching sensors and control logic.
ArduinoUltrasonic SensorsControl Logic

Rocker-Bogie Mining Robot (Final Year Project)

Lead Designerpersonal
An autonomous mobile robot for mine exploration and hazardous environment monitoring, using RGB vision-based mapping and gas sensors to replace expensive LiDAR systems. Navigates rough terrain to detect methane and carbon monoxide in real-time.
Mechanical DesignEmbedded ControlSensorsRGB Vision MappingAI PerceptionGas Sensors (Methane, CO)

Education

Bachelor of Engineering

National University of Sciences and Technology (NUST)Mechanical Engineering2022 – 2026

Coursework

Control Engineering & InstrumentationElectrical & Electronics EngineeringMechanical Design & ModelingKinematics and Dynamics

Societies & activities

Team ALIF (Autonomous Racing)Hack Club NUSTIMechESCS

Recognition

Certifications

Computer Vision & AI Training Program

Islamabad

Deep Learning & Computer Vision

Andrew Ng; Hugging Face

Governance & advisory

Liaison Director

Hack Club NUSTother

Deputy Director Finance

IMechE & SCSother
Hire MuhammadReach out about a role, a contract or a conversation.For recruiters

Muhammad's twin is AI, it can make mistakes.

More people on Kanvis

Browse more peopleCreate your own Kanvis