Welcome!

I'm Hynek Kydlíček, a passionate software developer with over four years of professional experience. Currently, I am pursuing my studies at MFF UK, where I continue to expand my knowledge and skills.

My interests

My expertise extends across various programming languages and paradigms, empowering me to tackle diverse software development challenges. My main interests is Artifical Inteligence, especially the areas of Natural Language Processing and Voice Synthesis.

Furthermore, I am interested in Distributed Systems and Cloud Computing.

Skills

Machine Learning

MLops

Since, I am Software Engineer, I have a great experience with CI/CD pipelines, mostly using Gitlab CI and Docker. For tracking I mostly used Weights and Biases, but I am familiar with MLflow as well. For data processing I usually create a custom pipeline depending on the framework or Apache Spark.

Deep Learning

I have a great experience with Tensorflow and Pytorch. I have used Pytorch in various projects, while the Tensorflow/Keras usage was just for a Deep Learning course. Apart from that I am well familiar with Pytorch-Lightning and Huggingface libraries. Other than that, I have also great experrience with scikit-learn and other for classical ML.

Web Development

Frontend

Unsuprisngly, I am familar with core of the web development, including HTML, CSS and Javascript. Considering the current trends, I at one points use React/Vue.js/Blazor, but I am definitely not the expert in the first two. The only framework, out of these 3 I enjoy using is React, which I used for this website. Blazor is terribly slow and horrible to use. Vue.js is not bad, but I am not a fan of the syntax.

Backend

I used to work with Django/Flask, but my hate for python has grown so much, that I switched to Node.js. That allows me to use Typescript, which I am a big fan of. I typically use Express.js, but I guess I could use Nest.js as well. Doesn't matter that much in Sass world.

Other

Linux

I have been using Linux for over 5 years. While I am by no means a kernel hacker, I am very familiar with the system and I able operatre it with ease, including the shell scripting.

AWS/Azure

I am mostly familiar with AWS, but I have also some experience with Azure. I have used AWS for deployment of 2 projects, including this website. I used a AWS SageMaker for training of my models, nowdays I mostly use lambda. This is the area, I would like to improve the most. I aim to learn terraform and kubernetes in the near future.

Projects

Czech News Classifier

Czech News Classifier is a project I developed for my bachelor thesis. It is a web application, which classifies czech news into 25 topics, guesses author's gender, day of the week and publication server. It is was trained on over 1M News articles using state-of-the-art NLP techniques. The project was trained using Pytorch-Lightning and Huggingface libraries. For logging and tracking I used Weights and Biases. The project was deployed using Gradio.

Technologies: python, Pytorch, Pytorch-Lightning, Huggingface, Weights and Biases, Gradio, Hydra

Website: https://huggingface.co/spaces/hynky/News-Classification

GitHub: https://github.com/hynky1999/Bakalarka-code

CZE-NEC

Czech News Dataset (CZE-NEC) is dataset consiting over 1.6M labeled czech news articles. To my knowledge this it the most complete news articles dataset while also one of the largest czech NLP datasets. Due to regulations, I am not allowed to share the dataset, but I am allowed to share the code I used to create it. Thus you can download it yourself :)

Technologies: Python, CmonCrawl, Docker

Website: https://huggingface.co/spaces/hynky/News-Classification

GitHub: https://github.com/hynky1999/Bakalarka-code

TranslateIT

TranslateIT, is very simple chrome extension, which I created because I couldn't find any good. It can translate selected text using DeepL. It also supports hotkeys and importantnly is incredibly fast and easy to use. It also support TTS in the target language. It is written in Typescript and uses React for the UI.

Technologies: Typescript, React, Chrome-extension

Website: Waiting for approval

GitHub: https://github.com/hynky1999/TranslateIT

Edulastic

I helped develop a backend for Edulastic with Snapwiz team. I was solely responsible for the creaation of the math evalution engine, which was used for validation of the student's answers. I also helped with the development of the other parts of the backend, including the API controllers

Technologies: Python, Antlr, Django, Postgresql

Website: https://www.edulastic.com/

CmonCrawl

Cmoncrawl is a tool for interaction and extraction from CommonCrawl, which I developed for my bachelor thesis. Since I develoeped it with distributivity and realiability in mind, I managed to process over 30M+ pages in a matter of hours, employing over 60 cores on Slurm cluster.

Technologies: Python, BeautifulSoup, Slurm, Apache Artemis, Docker

Website: https://pypi.org/project/CmonCrawl/

GitHub: https://github.com/hynky1999/CmonCrawl

Washing Manager

Washing Manager is a reservation system for washing machines, which I developed for a dormitory. It is a fullstack application written in Blazor, withe help of ASP.NET Core and PostgreSQL. It tought me that Blazor is not production ready and that internalization and dates are incredibly hard to get right.

Technologies: C#, Blazor, ASP.NET Core, PostgreSQL, Docker

GitHub: https://github.com/hynky1999/WashingManager