class AhmedSelim: def __init__(self): self.role = "Machine Learning Engineer" self.domains = ["ML/DL", "Computer Vision", "Robotics", "Control Systems", "Mechatronics"] self.education = "BSc Mechatronics & Robotics — GUC (GPA 3.4/4.0, Thesis: A+)" self.experience = "3+ years deploying ML models on robotics platforms" self.cloud = ["AWS SageMaker", "AWS Lambda", "S3", "EC2"] self.languages = ["English (Fluent)", "German (Intermediate)", "French (Intermediate)"] self.location = "Cairo, Egypt — works remotely with global teams" def now(self): return "Fine-tuning LLMs · deploying vision models · building autonomous systems"
| 🤖 ML & Deep Learning | 🦾 Robotics & ROS2 | ⚙️ Mechatronics | 🎛️ Control Systems |
|---|---|---|---|
| Fine-tuning LLMs & foundational models | Autonomous motion planning | Embedded sensing & hardware integration | MPC · EKF · PID |
| Vision models & object detection | Real-time ML on robotic platforms | Sensor fusion & prototype engineering | State machines · vehicle dynamics |
| Data pipelines & annotation workflows | CARLA · Gazebo simulation | Actuator & motor control | Kalman filtering · localization |
| AWS SageMaker deployment | ROS2 pub/sub · services · actions | PCB & system architecture | A* · Hybrid A* path planning |
PyTorch HuggingFace OpenCV scikit-learn NumPy
AWS SageMaker Docker Kubernetes GitHub Actions
ROS2 Linux Raspberry Pi Simulink
| Project | Description | Stack |
|---|---|---|
| 🔍 vision-motion-detection | Real-time motion detection, multi-object tracking & monocular depth estimation on embedded hardware | PyTorch OpenCV MiDaS Docker |
| 🚗 av-motion-planning-ros2 | Modular autonomous driving stack — A*, EKF, behavior FSM — validated in CARLA simulator | ROS2 Python CARLA EKF |
| ☁️ sagemaker-image-classifier | End-to-end fine-tuning pipeline (ResNet/EfficientNet) with SageMaker training jobs & live endpoint | AWS SageMaker PyTorch S3 |
| 🏷️ data-annotation-pipeline | Automated dataset curation — YOLOv8 pre-labeling, quality filtering, Label Studio, versioned S3 | YOLOv8 Label Studio S3 |
| 🤖 llm-document-processor | Mistral-7B fine-tuned via QLoRA for document classification & extraction on AWS Lambda | Mistral-7B QLoRA Lambda PEFT |
Building at the intersection of intelligence and physical systems Cairo, Egypt · Open to remote opportunities worldwide