Faran Ahmed Yasir

• Developed a vehicle tracking system across multiple cameras for theft protection and fraud detection, demonstrating expertise in machine learning and computer vision using Python.
• Successfully collected and annotated a large dataset of vehicle types and attributes, including jerry cans and taxi signals, and implemented active learning and cross-entropy loss on CUDA-powered machines.
• Overcame the challenge of vehicle ID switching among different cameras by utilizing techniques such as multi-camera frames, adjacent vehicles, and traffic rush, significantly improving the accuracy.
• Reduced clustering inter-camera associations by removing false detections of adjacent tracked vehicles, further enhancing the system's reliability.
• Trained and tested various YOLO models, including YOLOv3, YOLOv4, and YOLOv5, with different combinations for optimal accuracy, finally using FairMOT and YOLOv5 for detection and integration into a C# desktop application.
• Applied computer vision and machine learning techniques to enforce a dress code for individuals, developing a system that could impose charges for not following the appropriate dress code.
• Collected a large dataset of over 12, 000 images of clothing items, including Emirati clothing, and trained models for 12 clothing categories, including shirts, pants, bags, glasses, and thobes, utilizing YOLOv4 and YOLOv4 Tiny.
• Improved image quality by implementing the CLAHE preprocessing method, further enhancing the accuracy of the models.
• Collaborated with iOS/Android development teams to create an attendance system using AWS Rekognition service, demonstrating effective communication and teamwork skills, and completing the project in a CI/CD environment.
• Demonstrated expertise in software engineering fundamentals by integrating software solutions and deploying models on Azure.

Python, Machine Learning, Computer Vision, Azure, Aws, MySQL, Java, C#, OpenCV, Tensorflow, PyTorch, Git, CI/CD, Django, Spring Boot, Vehicle Tracking, Person Tracking, Fraudulent Activities Detection, Clothing Style Detection, Face Attendance, HR Portal

• Developed a lead generation and recommendation system that streamlines recruitment and delivers exceptional talent using Python.
• Utilized Linear SVM for accurate resume classification, reducing consultant workload and enabling efficient decision-making.
• Spearheaded meticulous data pre-processing to ensure pristine data quality and unlock valuable insights.
• Employed Gensim to summarize job descriptions and resumes, presenting results in a captivating and user-friendly format.

Python, Machine Learning, Natural Language Processing, Lead Generation System, Recommendation System, Profile Matchmaking, Git, Jira, Confluence.

Developed a lead generation and recommendation system that streamlines recruitment and delivers exceptional talent using Python.
Utilized Linear SVM for accurate resume classification, reducing consultant workload and enabling efficient decision-making.
Spearheaded meticulous data pre-processing to ensure pristine data quality and unlock valuable insights.
Employed Gensim to summarize job descriptions and resumes, presenting results in a captivating and user-friendly format.

Developed a groundbreaking fintech micro-influencer recommendation system, transforming how financial institutions connect with influencers for social media promotions using Python.
Implemented the churn prediction system using XGBoost to enhance customer retention efforts.
Implemented NLP techniques to ensure high-quality influencer recommendations, filtering out fake accounts and enhancing accuracy.
Utilized DeepFace and gender guesser libraries for gender and age identification of micro-influencers, enabling targeted marketing campaigns.
Applied VGG16 for image analysis enhancing influencer profile categorization and recommendation.
Integrated AWS services, deployed on VMs, and implemented Ethereum smart contracts with plasma integration for improved performance in the Mirian decentralized application.

Developed a lead generation and recommendation system that streamlines recruitment and delivers exceptional talent using Python.
Utilized Linear SVM for accurate resume classification, reducing consultant workload and enabling efficient decision-making.
Spearheaded meticulous data pre-processing to ensure pristine data quality and unlock valuable insights.
Employed Gensim to summarize job descriptions and resumes, presenting results in a captivating and user-friendly format.

Developed a vehicle detection system across multiple cameras for theft protection and fraud detection, demonstrating expertise in machine learning and computer vision using Python.
Successfully collected and annotated a large dataset of vehicle types and attributes, including jerry cans and taxi signals, and implemented active learning and cross-entropy loss on CUDA-powered machines.
Overcame the challenge of vehicle ID switching among different cameras by utilizing techniques such as multi-camera frames, adjacent vehicles, and traffic rush, significantly improving the accuracy.
Reduced clustering inter-camera associations by removing false detections of adjacent tracked vehicles, further enhancing the system's reliability.
Trained and tested various YOLO models, including YOLOv3, YOLOv4, and YOLOv5, with different combinations for optimal accuracy, finally using FairMOT and YOLOv5 for detection and integration into a C# desktop application.
Applied computer vision and machine learning techniques to enforce a dress code for individuals, developing a system that could impose charges for not following the appropriate dress code.
Collected a large dataset of over 12,000 images of clothing items, including Emirati clothing, and trained models for 12 clothing categories, including shirts, pants, bags, glasses, and thobes, utilizing YOLOv4 and YOLOv4 Tiny.
Improved image quality by implementing the CLAHE preprocessing method, further enhancing the accuracy of the models.
Collaborated with iOS/Android development teams to create an attendance system using AWS Rekognition service, demonstrating effective communication and teamwork skills, and completing the project in a CI/CD environment.
Demonstrated expertise in software engineering fundamentals by integrating software solutions and deploying models on Azure.

Developed a groundbreaking fintech micro-influencer recommendation system, transforming how financial institutions connect with influencers for social media promotions using Python.
Implemented the churn prediction system using XGBoost to enhance customer retention efforts.
Implemented NLP techniques to ensure high-quality influencer recommendations, filtering out fake accounts and enhancing accuracy.
Utilized DeepFace and gender guesser libraries for gender and age identification of micro-influencers, enabling targeted marketing campaigns.
Applied VGG16 for image analysis enhancing influencer profile categorization and recommendation.
Integrated AWS services, deployed on VMs, and implemented Ethereum smart contracts with plasma integration for improved performance in the Mirian decentralized application.

Developed a vehicle detection system across multiple cameras for theft protection and fraud detection, demonstrating expertise in machine learning and computer vision using Python.
Successfully collected and annotated a large dataset of vehicle types and attributes, including jerry cans and taxi signals, and implemented active learning and cross-entropy loss on CUDA-powered machines.
Overcame the challenge of vehicle ID switching among different cameras by utilizing techniques such as multi-camera frames, adjacent vehicles, and traffic rush, significantly improving the accuracy.
Reduced clustering inter-camera associations by removing false detections of adjacent tracked vehicles, further enhancing the system's reliability.
Trained and tested various YOLO models, including YOLOv3, YOLOv4, and YOLOv5, with different combinations for optimal accuracy, finally using FairMOT and YOLOv5 for detection and integration into a C# desktop application.
Applied computer vision and machine learning techniques to enforce a dress code for individuals, developing a system that could impose charges for not following the appropriate dress code.
Collected a large dataset of over 12,000 images of clothing items, including Emirati clothing, and trained models for 12 clothing categories, including shirts, pants, bags, glasses, and thobes, utilizing YOLOv4 and YOLOv4 Tiny.
Improved image quality by implementing the CLAHE preprocessing method, further enhancing the accuracy of the models.
Collaborated with iOS/Android development teams to create an attendance system using AWS Rekognition service, demonstrating effective communication and teamwork skills, and completing the project in a CI/CD environment.
Demonstrated expertise in software engineering fundamentals by integrating software solutions and deploying models on Azure.