06.09.2022 aktualisiert
100 % verfügbar
Software Developer, Machine / Deep Learning Engineer
Bayreuth, Deutschland
Deutschland
M.Sc. Computer Science (Data Science)Skills
Keras PyTorchPythonVisual-Studio C#.netAPIs GITAWS (Amazon WebServices)MySQLMSSQLMachine Learning, Data, Python, relationale DatenbMachine LearningDeep Learning mit Tensorflow
I have the following skills, experience, and knowledge:
- C#, Entity Framework, .NetCore, MVC, MSSQL, Web-Forms, API
- PHP, Larvel, MySQL, JavaScript, CSS, HTML, Bootstrap
- Machine Learning and Deep Learning
- Keras, Scikit-learn, TensorFlow, Pytorch
- ANNs, CNNs, GANs, and NLP
- Data augmentation methods
- Numpy, Pandas, Matplotlib, SciPy, Seaborn, etc.
- Data pre-processing methods
Sprachen
DeutschgutEnglischverhandlungssicher
Projekthistorie
Master Research Thesis Grade (1,0)
Convolutional Neural Networks Based Transfer Learning for Image Classification
Convolutional Neural Networks Based Transfer Learning for Image Classification
- Implemented in Python using Keras API with TensorFlow backend
- Designed 9 transfer learning models (ImageNet CNNs) including a novel VGG16BN
- Used two model-level approaches: network and parameter-based transfer learning
- Evaluated models on three publicly available datasets (scene-centric: MIT-Indoor-67, object-centric: Fashion-MNIST & Event-8)
- Improved generalization and reduced overfitting problem through affine transformations and Random Erasing data augmentation
- Evaluated performance affecting factors such as image pre-processing and image resolution
- Improved convergence of models through callbacks
- Analyzed performance of transfer learning models on different nature of datasets
- Prototype of an intelligent platform to offer discounted items in the local areas
- Registering the businesses to list discounted items
- Personalized users’ screens based on their preferences and most bought items
- Integrating Google map to locate registered shops
- Integrating Barcode reader to show the price of items and to the checkout payment process
- PHP API to perform CRUD operations
- Performance evaluation of Intel Arria 10GX FPGA accelerator through Intel SDK FPGA for OpenCL
- Utilizing maximum resources of an accelerator to increase throughput(GFLOPS)
- Implemented Naïve and Loop-Tiling Matrix Multiplication
- Loop-Unrolling performance optimization
- Floating-Point operations optimization
- SIMD (automatic vectorization) optimization
- Analyzed the limitation of the compiler in the context of optimization