Enoch Kumi Asamoah

• Consulted with managers to determine and refine machine learning objectives.
• Selected appropriate datasets and data representation methods.
• Designed machine learning systems and self-running artificial intelligence (AI) software to automate predictive models.
• Developed ML algorithms to analyze huge volumes of historical data to make predictions.

1. Misbehaving Campus Detection System
Role: Trained machine learning models to detect student misbehavior (e.g., smoking) in real-time surveillance footage.
Contribution:
Curated and annotated datasets of student behaviors for model training.
Collaborated with developers to optimize the model's accuracy for diverse environments.
Impact: Improved campus monitoring and safety.

2. License Plate Recognition and Traffic Violation Detection
Role: Developed and annotated datasets for license plate recognition and traffic violation scenarios.
Contribution:
Focused on diverse lighting, weather, and motion conditions for robust training.
Provided feedback on model performance based on real-world test data.
Impact: Enabled efficient traffic law enforcement with high accuracy.

3. Plant Disease Detection Using ML
Role: Trained models to identify diseased plant leaves from large agricultural datasets.
Contribution:
Labeled extensive datasets of healthy and diseased leaves.
Worked with data augmentation techniques to enhance model performance.
Impact: Empowered farmers with early detection tools, reducing crop loss.

4. LiDAR 3D Environment Segmentation
Role: Provided labeled point cloud data for training LiDAR-based segmentation models.
Contribution:
Assisted in categorizing objects like pedestrians, vehicles, and buildings.
Ensured consistent annotations for improved model accuracy.
Impact: Supported the development of autonomous navigation systems.

5. NSFW Content Blocking for School Systems
Role: Trained models to detect and block NSFW content searched by students in a school environment.
Contribution:
Built and curated datasets of NSFW and safe content for training.
Tested and fine-tuned the model for high precision and low false positives.
Impact: Ensured a safer and more focused online experience for students.