After a few months of work, I've released the first stable version of a filesystem benchmark that is able to simulate the aging of filesystems and can be used to test filesystems for their performance degradation over time.
The project is open-sourced thanks to Red Hat's open source policy and can be found on the following GitHub repository:
https://github.com/janjurca/Filestorm C++FilesystemsPackage Management
5/2023
Even though my business was not successful, machine learning remained my passion and hobby. I've started working on a new project called Meteonet, which I later open-sourced in the following GitHub repository:
https://github.com/janjurca/meteonet
The model uses a visual transformer to predict the weather based on satellite images.
Along with the model, I've also open-sourced a tool for easy training and deployment, and a script for global weather prediction. Machine learningPVDMViTDiffusion modelssr-gan
11/2023
Red Hat: Full-time
Even during a hiring freeze, I managed to secure a full-time position at Red Hat.
In addition to my previously mentioned responsibilities, I also started managing our hardware lab and began working on a new open-source filesystem aging benchmark. C++LinuxDeep Filesystem KnowledgeOpen Source
10/2023
Zoner: Machine Learning Consultant
I was approached by Zoner to help them kickstart their machine learning projects. In the three months I was able to work with them, I helped set up their machine learning infrastructure, introduced relevant tools and machine learning projects aligned with their business goals. Machine LearningConsultingKubernetesmicrok8sMLflowSeldon CoreStable DiffusionImage-Related ML Tasks and Models
7/2023
Even though I had a very well-trained model for speech recognition and diarization that outperformed state-of-the-art models for the Czech language, and I had a functional speech synthesis model with speech cloning capabilities, I started to slowly deprioritize it due to the increasing competition in the field and my lack of business and marketing skills.
7/2023
Started working on my own project focused on machine learning in speech recognition and synthesis. Machine LearningSpeech RecognitionSpeech SynthesiswandbPyTorchMLflowwav2vecWhisperYourTTSVall-e
1/2023
Decided to end my PhD studies and focus on my career in the industry.
Submitted a paper to the MEDICON 2023 conference as the first author.
Developed a deep learning model based on the V-Net architecture for automatic inference of the radiological planes of the heart from 3D MRI sequences. This model improves the accuracy and efficiency of medical imaging by generating heatmaps of probable plane positions, enabling faster and more effective diagnosis and treatment. Demonstrated the network's capability to accurately locate cardiac landmarks, even with anisotropic voxel data.
Springer link Project GithubMachine LearningFirst Author Conference Paper
1/2023
Completely reworked (innovation) and lectured lab exercises for the course "Biomedical Data Visualization". Course link C/C++OpenGL
Faculty of Electrical Engineering and Communication, Doctoral study Biomedical Technology and Bioinformatics
Machine learning in tomography image processing BioengineeringMachine LearningMedical ImagingTomography
The thesis focused on developing a web application to evaluate and recommend primer pairs for 16S rRNA amplicon sequencing based on user-specific needs, incorporating a database created through extensive data mining. The project also optimized the analysis algorithm for performance, enabling users to submit their own primer sequences and receive comprehensive evaluation results, including amplified region positions, sensitivity, and specificity.
I was awarded the Dean's Award for this work. Dean's AwardData MiningLarge Database Speed OptimizationDjangoCeleryWebsocketKubernetesDockerPostgreSQLRedisBioinformaticsMetagenomics
6/2022
Red Hat Promotion: Software Quality Engineer
I created a machine learning model for automatic classification of performance test results, which are by their nature very noisy and thus impossible to classify by simple rules. Additionally, I advocated for the use of OpenShift for deploying many internal tools. PythonMachine Learning (scikit-learn, PyTorch, NumPy, Pandas)LinuxDjangoAnsibleOpenShiftTest Automation
6/2020
Faculty of Information Technology, Master's Degree Program: Bioinformatics and Artificial Intelligence
Developed a neural network for pedestrian identification from video sequences, incorporating person, face, and gait recognition. Utilized pretrained networks for person and face recognition, and implemented various networks for gait recognition, comparing their performance. Created a custom dataset and tools for its almost automatic generation, culminating in a multimodal fusion approach using neural networks for final pedestrian recognition. PytorchMachine learningDataset creation
6/2019
Red Hat Promotion - Associate Software Quality Engineer - Kernel Performance Team
Received additional responsibilities within the Kernel Performance Team, focusing on the deployment of various internal automation tools and web applications using Django. PythonLinuxDjangoAnsibleTest Automation
7/2018
Redhat internship - Kernel performance team
Storage and filesystem Performance tests automation. PythonLinuxTest automation
8/2017
Participation in the project "Named entity recognition and relation extraction for Decipher" of group KNOT.
Automating the execution of scripts extracting information for the knowledge base. Academic cooperationPythonWeb scraping
6/2017
Faculty of Information Technology , Bachelor's degree study program: Information Technology
2016-2019ProgrammingMachine learning basicsFundamentals of computer scienceDatabasesNetworkingOperating systemsAnd more...
9/2016
Mathias Lerch Gymnasium, Brno High school2012-2016
After a few months of work, I've released the first stable version of a filesystem benchmark that is able to simulate the aging of filesystems and can be used to test filesystems for their performance degradation over time. The project is open-sourced thanks to Red Hat's open source policy
I've trained a ViT model capable of nowcasting weather radar images for several hours in future. The model uses a visual transformer to predict the weather based on satellite images. Along with the model, I've also open-sourced a tool for easy training and deployment, and a script for global weather prediction. Please see the GitHub repository for more information.
Machine LearningPythonPyTorchTomographyComputer visionConference paper
Submitted a paper to the MEDICON 2023 conference as the first author. Developed a deep learning model based on the V-Net architecture for automatic inference of the radiological planes of the heart from 3D MRI sequences. This model improves the accuracy and efficiency of medical imaging by generating heatmaps of probable plane positions, enabling faster and more effective diagnosis and treatment. Demonstrated the network's capability to accurately locate cardiac landmarks, even with anisotropic voxel data.
As a school project we have been tasked to train a Generative Adversarial Network on fingerprint images. Trained network should be later used to generate synthetic fingerprint images and break the fingerprint recognition system. The project was done in Python and the neural network was implemented in PyTorch. We have been able to combine trained network with genetic algorithm to generate such synthetic fingerprint image which was able to break the fingerprint recognition system.
As a school project we have been tasked to train a Dense passage retrieval on open domain question answering. Even though it was only a school project, I have invested a lot of time and effort into it, to learn as much as possible. The project was done in Python and the neural network was implemented in PyTorch. I've based the implementation and training on the original Facebook AI Research paper. I've conducted dimensionality reduction in order to discover the ability of the model to perform on a smaller vector space. The project was a great learning experience and I have been able to produce a interesting results.
As a school project we have been tasked to train a OCR neural network. Since it was only a school project, we haven't aspired to create a perfect OCR, but rather build the best we could in the given time frame. The project was done in Python and the neural network was implemented in PyTorch. We have used relatively small custom convolutional architecture, optimized for CTC loss. As dataset we used Brno Mobile OCR Dataset. Overall, the project was a great learning experience and we have been able to achieve a decent OCR accuracy.
Filestorm
After a few months of work, I've released the first stable version of a filesystem benchmark that is able to simulate the aging of filesystems and can be used to test filesystems for their performance degradation over time. The project is open-sourced thanks to Red Hat's open source policy
Date:
2020
Service:
Linux
Meteonet
I've trained a ViT model capable of nowcasting weather radar images for several hours in future. The model uses a visual transformer to predict the weather based on satellite images. Along with the model, I've also open-sourced a tool for easy training and deployment, and a script for global weather prediction. Please see the GitHub repository for more information.
Date:
2020
Service:
Machine Learning
Deep-Learning Based Automatic Determination of Cardiac Planes in Survey MRI Data
Submitted a paper to the MEDICON 2023 conference as the first author. Developed a deep learning model based on the V-Net architecture for automatic inference of the radiological planes of the heart from 3D MRI sequences. This model improves the accuracy and efficiency of medical imaging by generating heatmaps of probable plane positions, enabling faster and more effective diagnosis and treatment. Demonstrated the network's capability to accurately locate cardiac landmarks, even with anisotropic voxel data.
Date:
2020
Service:
Machine Learning
Fingerprint GAN
As a school project we have been tasked to train a Generative Adversarial Network on fingerprint images. Trained network should be later used to generate synthetic fingerprint images and break the fingerprint recognition system. The project was done in Python and the neural network was implemented in PyTorch. We have been able to combine trained network with genetic algorithm to generate such synthetic fingerprint image which was able to break the fingerprint recognition system.
Date:
2020
Service:
Machine Learning
DPR on ODQA
As a school project we have been tasked to train a Dense passage retrieval on open domain question answering. Even though it was only a school project, I have invested a lot of time and effort into it, to learn as much as possible. The project was done in Python and the neural network was implemented in PyTorch. I've based the implementation and training on the original Facebook AI Research paper. I've conducted dimensionality reduction in order to discover the ability of the model to perform on a smaller vector space. The project was a great learning experience and I have been able to produce a interesting results.
Date:
2020
Service:
Machine Learning
OCR
As a school project we have been tasked to train a OCR neural network. Since it was only a school project, we haven't aspired to create a perfect OCR, but rather build the best we could in the given time frame. The project was done in Python and the neural network was implemented in PyTorch. We have used relatively small custom convolutional architecture, optimized for CTC loss. As dataset we used Brno Mobile OCR Dataset. Overall, the project was a great learning experience and we have been able to achieve a decent OCR accuracy.
