- Developed a Facial Recognition System using MTCNN for detection, FaceNet for recognition, and embedding for scalable database storage, achieving 97% accuracy. Deployed with Flask APIs and TensorFlow Serving.- Implemented Forklift detection using MobileNetV2, converted to TFLite and 4-bit quantized for Google Coral Edge TPU, achieving 95% accuracy and reducing accidents by 90%. Deployed with Docker and Kubernetes.- Created a Saudi license plate recognition system using YOLOv5 for plate detection and custom CNN models for recognition, achieving 98% accuracy.- Developed a crowd counting and tracking solution using P2PNet with large-scale deployment via Tf-serving, Kubernetes, and Docker.- Integrated GenAI technologies using models like GPT-3, GPT-4, LLaMA, LLaMA Index, and Gemini, enhancing data processing and decision-making capabilities. Utilized vector databases like Qdrant, Pinecone, and Weaviate for efficient data retrieval and storage.- Applied NLP techniques with Whisper, Flask APIs, FastAPI, Django, and BiLSTM, improving text analysis and information retrieval.- Architected solutions for complex business problems involving large-scale real-time and asynchronous applications.

As a freelance Data Science and Machine Learning Engineer, I provided global clients with advanced tools and solutions, leading to significant business improvements.- Developed machine learning models like Random Forest, XGBoost, Linear Regression etc.- Implemented data analytics solutions to improve decision-making processes and operational efficiency.- Created NLP models using BERT, Bidirectional LSTM and GPT2 etc for enhanced text analysis.- Built computer vision apps with CNN's, YOLO family, MaskRCNN and ResNet, DeepSort, for various applications.- Delivered projects on time, achieving high client satisfaction.- Conducted training sessions and workshops to boost client teams' data science capabilities.

- Designed and implemented an Arduino-based door entry system using Tiny YOLO-V5 for real-time face detection, resulting in a 40% improvement in security efficiency.- Developed a real-time mask detection and person counting system for healthcare applications using MobileNetV2 and DeepSort, achieving a 95% accuracy rate and enhancing monitoring capabilities by 30%.- Built comprehensive dashboards, ETL pipelines, and workflows using ReactJS+Django, increasing data ingestion and processing efficiency by 50%.

It was a pleasure being part of the Microsoft Learn Student Ambassadors program. This global initiative provided on-campus ambassadors with opportunities to lead in their local tech communities and develop critical technical and career skills through tools and mentorship from industry experts.- Delivered workshops on AI topics like OpenCV and facial gesture recognition.- Organized hackathons to promote hands-on learning and innovation.- Mentored students on object recognition techniques and AI project development.- Conducted sessions on the basics and applications of AI, including vector databases.- Encouraged students to explore and pursue careers in AI, fostering a collaborative learning environment.