Hermina is available for hire
Hire HerminaHermina Petric Maretić
Verified Expert in Engineering
Software Developer
Location
Zagreb, Croatia
Toptal Member Since
August 27, 2014
Hermina is a developer with proven skills in data mining, machine learning, and mathematical optimization. When building a project, she gives special attention to algorithm efficiency, putting an emphasis on creating quick and optimized software.
Portfolio
Mozgalo Competition
Bootstrap, JavaScript, PHP, MySQL, Django, Python
University of Zagreb
C#, SAS, R, HTML, JavaScript, SQL, PHP, MATLAB, Python, C++, C
Experience
Availability
Part-time
Preferred Environment
MATLAB, C++, Django, Python, Windows, Linux
The most amazing...
...thing I've coded is a hybrid of an ants algorithm for optimization. It achieves results in seconds that take other heuristic algorithms days to reach.
Work Experience
Competitor
2014 - 2014
Mozgalo Competition
- Created a web scraper that gathered millions of sentences describing beers, using PHP.
- Determined which aspect of beer a sentence is discussing, using SVD in Python.
- Graded sentences with machine learning algorithms and determined the final grade for every beer.
- Developed a web application to present an extensive analysis of results using Bootstrap, HTML, JavaScript, Python, and Django.
- Built a recommender system using Python.
Technologies: Bootstrap, JavaScript, PHP, MySQL, Django, Python
Student
2009 - 2014
University of Zagreb
- Developed face recognition software using Singular Value Decomposition (SVD) in MATLAB.
- Created text classification software thoroughly analyzed approaches using Python.
- Built a student web page using PHP, JavaScript, MySQL, CodeIgniter, and HTML.
- Created a metaheuristical approach to the Hamiltonian completion problem using MATLAB.
- Implemented an algorithm for finding closest points. Used an object-oriented paradigm, implemented data structures, and competed in algorithm competitions using C++.
- Performed statistical analysis of data in R and SAS.
Technologies: C#, SAS, R, HTML, JavaScript, SQL, PHP, MATLAB, Python, C++, C
Researcher (Intern)
2013 - 2013
EPFL (Ecole polytechnique fédérale de Lausanne)
- Conducted research in Bioinformatics and Computational Biology.
- Implemented an algorithm for finding the shortest path in a graph using a specific set of biology-conditioned rules.
- Worked on an iterative algorithm to improve evolutionary trajectories.
- Showed that a simplified model with only DCJ (double-cut-and-join) operations wouldn't work if transposition, deletion, and duplication operations were added.
- Worked on a modification of the model allowing it to handle all of the above-mentioned operations.
Technologies: Linux
Developer (Intern)
2012 - 2012
Ericsson Nikola Tesla
- Redirected calls using Java.
- Created a scalable database for users.
- Created a dynamic web page using Bootstrap.
- Tested the page in a real-life active telecommunication network.
- Showed user location using Google Maps.
Technologies: JavaScript, Bootstrap, Google Maps API, Java
Experience
Mozgalo
A data mining project for finding the best beer. Information is gathered from web pages, stored in a database, and analyzed using numerous machine learning methods. The application is realized as a dynamic web page with numerous analysis options: best beer by category, by city, popularity over time, and more. In addition, a recommender system was created with the possibility of grading a user's own comments, as well as options to renew the data and explore the database. The project was realized in Python, Django, PHP, and JavaScript.
Text classification by subjectivity
https://github.com/Hermina/Machine-learningIt classifies text using a naive Bayes classificator and support vector machine classificator. It provides extensive analysis on best feature selection and achieves very good results (92% accuracy). Technologies used: Python, NLTK, SciPy, and NumPy.
Student website
A simple website for students. Carefully chosen features combine everything a student needs. It was realized in PHP, CodeIgniter, Bootstrap, and JavaScript.
Metaheuristics
https://github.com/antegrbic/metaheuristicsA metaheuristical approach to the Hamiltonian completion problem. It features genetic algorithms, an Ants algorithm, an immunological algorithm, and a hybrid of an Ants algorithm, which gave the best results by far. It also features extensive analysis of the solution approaches, and was implemented in MATLAB.
Skills
Languages
Python, C++, C, SQL, PHP, R, JavaScript, C#, SAS, Java, HTML
Tools
LaTeX, MATLAB
Paradigms
Object-oriented Design (OOD), Object-oriented Programming (OOP), Test-driven Development (TDD)
Platforms
Linux, Windows
Other
Machine Learning
Frameworks
Bootstrap, Django, .NET
Libraries/APIs
SciPy, NumPy, Google Maps API, Matplotlib
Storage
MySQL, PostgreSQL
Education
2013 - 2015
Master's Degree in Mathematical Statistics
University of Zagreb - Zagreb, Croatia
2012 - 2014
Master's Degree in Computer Science
University of Zagreb - Zagreb, Croatia
2009 - 2012
Bachelor's Degree in Mathematics
University of Zagreb - Zagreb, Croatia