Ashhad Ahsan Rehman Email & Phone Number

Koblenz, Rp, De

Karlsruhe, Baden-Württemberg, Germany

Fine-Tuning Meta Segment Anything Model for German Number Plate RecognitionModel Customization: Led the fine-tuning of the Meta Segment Anything Model (SAM) to enhance its accuracy and robustness specifically for recognizing German number plates. This involved adapting the model to handle the unique alphanumeric structure and formatting of German vehicle registration plates.Data Preparation & Annotation:Curated and pre-processed a comprehensive dataset of German number plates, including manual annotation and augmentation techniques to ensure the model was trained on a diverse set of real-world scenarios.Model Training & Optimization: Conducted multiple iterations of model training, implementing hyperparameter tuning and optimization strategies to maximize model performance, particularly focusing on minimizing false positives and improving detection in various lighting and weather conditions.Model Evaluation & Validation: Performed rigorous evaluation of the fine-tuned model using a combination of cross-validation and real-world testing datasets. Benchmarked the model against industry standards and ensured its readiness for deployment.Research & Development: Engaged in extensive literature review and analysis of state-of-the-art methods in number plate recognition and segmentation models. Applied insights from recent research papers to refine the approach and incorporate cutting-edge techniques into the model fine-tuning process.Documentation & Reporting: Documented the entire process, from data preparation to model evaluation, and presented findings and recommendations to stakeholders. Created detailed reports outlining the performance metrics and areas for future improvement.

Project Highlights: Sound Classification Algorithm: Developed and deployed a sound classification algorithm using Facebook's Wave2Vec2 on Raspberry Pi for predicting neighborhood crimes, achieving a model accuracy improvement from 70% to 92%. This project is currently in use across several private properties in the USA. GPT-4 and BERT Implementations:Leveraged GPT-4 and BERT for natural language processing tasks, including text generation, sentiment analysis, and entity recognition. Web Scraping and Data Analysis: Executed multiple web scraping projects to gather data from various sources for analysis and predictive modeling. Machine Learning and Deep Learning: Implemented machine learning algorithms and deep neural networks (DNN) for various clients, enhancing model performance and scalability.Technical Skills: Programming and Development: Python, Go, JavaScript, Node.js, Flask, FastAPI, R Data Engineering and Analysis: PySpark, ETL processes, data visualization, data mining, predictive modeling Machine Learning and AI:TensorFlow, Keras, PyTorch, BERT, GPT-3, GPT-4, deep learning, LLM, computer vision, natural language processing (NLP) Cloud Computing and Deployment:AWS (EC2, CLI, Lambda, SageMaker), Firebase, Cloud Firestore, Docker, Amazon Web Services (AWS) Database Management:NoSQL, MongoDB, SQL databases Web and API Development: RESTful Web Services, back-end web development, web applications Other Tools: Robotic Process Automation (RPA), ChatGPT, data structures, algorithms, optimizationCommunication and Collaboration: Worked closely with clients to understand their requirements and deliver high-quality solutions, ensuring effective communication and timely project completion.

Worked with top notch Google technology to create data science and machine learning applications for local merchants.

Master Of Science - Ms, Web And Data Science

Bachelor'S Degree, Software Engineering

Activities and Societies: Member of Google Developer Student Club