About Me

Anna Mrukwa

I am a Deep Learning Engineer with Computer Vision and a part of research team of Department of Data Science and Engineering at Silesian University of Technology. I focus on the Artificial Intelligence in Healthcare, especially on detecting cancer via several different approaches, depending on the tissue. However, I am trying to be involved in many other areas as well, hence my three year industrial work with road data processing.

My Career

Cufix PL

I was promoted to the Deep Learning Engineer and continued working on the computer vision related tasks.

Jun. 2023
Deep Learning Engineer

Silesian University of Technology

I started my Master studies on Silesian University of Technologies on interdiscplinary course, called Control, Electronic and Information Technology, with the major in Data Science. This helped me broaden the knowledge I used in the projects I worked on both in my industry job and in my research.

Mar. 2023
Master studies

Cufix PL

I started working at Cufix PL. I was involved in creating solutions for object detection for autonomous vehicles as well as camera systems. We also created Polish blog on computer vision topics.

Apr. 2021
Junior Deep Learning Engineer

Silesian University of Technology

I started my Bachelor studies on Silesian University of Technologies on interdiscplinary course, called Control, Electronic and Information Technology, with the major in Informatics. The studies were held in English and during that time I developed versatile skills I use to this day.

Oct. 2019
Bachelor studies

My Skills

My Projects

K-means in Python

I strived to follow the scikit-learn pipelines during the algorithm implementation. To make visualization possible, I also implemented PCA algorithm. I programmed different initialization methods, as well as Dunn Index to find the optimal number of clusters. To validate the results, I wrote k-cross validation.

K-means in C++

For this project, I started with designing the ML canvas, which helped me define the scope of the project. I implemented most of the necessary classes by myself, letting the user choose the initialization method, distance metrics, index for finding the optimal number of classes. I also implemented the PCA algorithm alongside the k-means. The whole code was consistently unit tested.

Shortest Path Search

The project is an implementation of Dijkstra algorithm in C++. The whole program can be invoked from the command line, providing additional guidance to the user. Moreover, I automatized the unit tests, as well as releases, with the usage of custom Github Actions via workflows I prepared.

Spam detection

In this project I built a simple spam detection application for e-mails using a SVM model and Bag of Words. Then I deployed it with Flask and Docker.

SAUCIE

This is the project I did for my Bachelor thesis. I converted the model dependencies to tensorflow 2, performed unit tests, checked results reproducibility, compared them with other popular methods via a series of Polyaxon experiments. The model is deployed using streamlit.

Mask detection

I built a simple Random Forest Classifier checking if on the face (previously detected by openCV DNN) there is a mask, correctly placed. The program marks the face as well as annotate the result. I deployed the model as Docker container with Flask. The app runs on images, as well as live stream video.


Projects in Progress

RCB compare

This project aims to compare two treatment methods based on their Residual Cancer Burden scores. I created the shiny web application and containerised it with Dockerfile using multistage build. Right now I am the administrator of the system.

Enrichment analysis

This is my Master thesis and the basis of some of my publications. I rewrote several SOTA Pathway Activity Scores from R and tested their compliance with the original. We introduced new GMM and k-means based thresholding methods for the heterogeneity analysis and VAE based PAS algorithm.