Muhammed Adil YATKIN

Non-Intrusive Monitoring (NILM), Development of customized AI models from scratch for smart house energy management system. ML model trainings and deployment through Vertex AI in Google Cloud. Research on recent publications. AI product development.

Currently, I am working in a project that is related with the fracture estimation on materials based on Neural Networks. Specifically my focus is on Customized Recurrent Neural Network Cells. The research includes multidisciplinary work which is connected to both Finite Element Methods and Machine Learning. I am flexible on using different programming languages, Python, Matlab, C, C++, Fortran.

3D object recognition, tracking. Radar + camera fusion in self-driving systems. Kalman filtering. Research on the state of the art in autonomous driving and rapid prototyping for product development. Training different machine learning models, hypertuning, deployment.

Object Detection, Multi-Object Tracking, Speed Estimation based on CCTV cameras. Data Analysis, Data Augmentation, Data Science research especially focuses on current state-of-the-art methods and applications on product development. Especially research publications that are related to the Computer Vision field. Training machine learning models in different frameworks such as PyTorch, TensorFlow, Darknet and deploy through TensorRT, ONNX, Deepstream etc. Carrying an evaluation tools to production environment by doing a deep research on different metrics which is related with different complex computer vision problems such as MOTA, HOTA, mAP etc. Developing some computer vision algorithms from scratch for fast innovative solutions.

Data augmentation, Synthetic Data Generation, Customized training on parking signs and deploying in production environment. The goal was extracting the parking signs map over all United State region by using the street level imagery. An end to end optic character recognition model as Fast Oriented Text Spotting Network has been investigated, fine tuned, customized and deployed in production environment. Master thesis with the Title "Deep Neural Networks based Optical Character Recognition Systems on Parking Signs". All product development has been done by me under the supervision of Dr. Till Quack, ETH Zurich, VP Product Mapillary

Mapillary is an important company working in the field of computer vision and trying to make a visual map of the world. I worked as a Computer Vision in the product team.

Specifically, my field was Optical Character Recognition. Supervisor Dr. Till Quack ETH Zurih.