I'm Marija, a girl born and raised in Belgrade, Serbia 🇷🇸, who fell in love with the STEM field at the age of 10. Chasing my dreams led me to Lausanne, Switzerland 🇨🇭, where I'm on adventure of gaining new knowledge and experiences!

I completed my BSc in Electrical Engineering and Computing ⚡💻 @ University of Belgrade, School of Electrical Engineering and I'm now studying towards MSc in Life Sciences Engineering with Minor in Data Science 🧬📊 @ EPFL. Currently, I'm mostly interested in ML/AI and its applications into biomedical field.

My GitHub page contains everything I've worked on during my Master's at EPFL, mainly cool projects and assignments that were part of my coursework.

Come take a look! 👀🚀

• Homework 1 - Pandas, Data Wrangling, and Data Visualization
• Homework 2 - Preprocessing data to make it amenable to statistical analysis and machine learning models. Training and testing out-of-the box machine learning models in sklearn. Carrying out simple multivariate regression analysis using statsmodel. Usage of propensity score machine to conduct causal analysis.
• Project - Analyzing beer preferences across U.S. states 🇺🇸🍻 and whether they can be linked to their political affiliation. Scored maximum 100/100 points and was placed among top 10 best projects 🏆 at the course.

Semester project @ Laboratory of Computational Biology and Theoretical Biophysics (Bitbol lab). Worked on capturing phylogenetic relationships using transformer-based deep learning architecture (MSA Transformer) and synthetic protein multiple sequence alignments (MSA). Built in python using esm, PyTorch and peft libraries and Cluster computing.

• Assignments - Covering various data science techniques especially for handling large scale real-world data.
• Project - Python based robust journey planner for the SBB (Swiss Federal Railways). The planner calculates the fastest routes between stops, providing a list of routes with confidence levels. Project was graded with maximum grade of 6.

Repository contains two assigments and mini project.
First assignment contains scientific analysis in R on phylogenetic relationships and population genetics.
Second assignment contains scientific analysis in R on bulk RNA sequencing, alternative splicing and single-cell RNA sequecing data.
Mini project contains reproduction of results from the paper Gene Expression Profile on Human Mesenchymal Stromal Cells Exposed to Hypoxic and Pseudohypoxic Preconditioning - an Analysis by RNA Sequencing accompanied with report and the code in R.

Repository contains the code for implementing DQN and Dyna Agents in Mountain Car enviroment as well as reproducing the results from the report. All the implementations are using torch library.

ML4Science project done in collaboration with Laboratory of the Physics of Biological System (LPBS) explores for mutants with enhanced protein fitness using AlphaFold cluster, AlphaFold, XGBoost, and BERT. Check out project pipeline in the repository.

• Assignments - Covering various deep learning concepts, from simple neural networks to GNNs and transformers.
• Project - Extending few-shot learning benchmark with RelationNet model.