Sandro Paradžik
sandro.paradzik@gmail.com • sandropa.github.io GitHub: @sandropa • LinkedIn: in/sandro-paradzik
(Recommended: Download PDF Version)
Education
BSc in Theoretical Computer Science | University of Sarajevo Expected Graduation: July 2025
Research Experience
CaCTüS Intern | MPI for Biological Cybernetics | Tübingen, July – September 2024
- Applied dynamic mode decomposition (DMD) to analyze calcium imaging data from zebrafish larvae brains, aiming for data-driven discovery of functional subnetworks.
- Extracted significant spatiotemporal patterns, notably oscillatory dynamics in the optic tectum potentially reflecting winner-take-all circuit mechanisms.
- Demonstrated DMD’s advantage in capturing coupled spatial and temporal neural dynamics compared to traditional methods (e.g., PCA, ICA, DFT).
- Contributed to research within the RoLi Lab (Drs. Drew Robson & Jennifer Li), supervised by Dr. Sophie Aimon.
- Selected for a competitive program with a 1.35% acceptance rate in 2024.
PSI Project Participant | ANNT | Sarajevo, July – December 2023
- Investigated uncertainty quantification techniques, including approximate Bayesian computation (ABC) and history matching (HM), for agent-based models in a UROP-like project.
- Critically reviewed scientific literature and presented findings on UQ approaches.
- Organized by Association for the Advancement of Science and Technology (ANNT). Supervised by Dr. Kenan Šehić.
Industry Experience
Math Expert | Mercor | October 2024 – Present
- Training LLM models (correcting their responses and providing solutions on which these models can be trained on) to solve math problems, covering high school to olympiad level problems.
AI/ML Intern | One Thousand | Berlin, February – April 2025
- Developed and maintained components for automating business document processing (e.g., invoices, offers) workflows using Python, FastAPI, and LLM APIs.
- Applied prompt engineering techniques for structured information extraction from documents.
- Contributed to short-term projects including predictive modeling for demand forecasting and automated web data gathering for lead generation support.
- Utilized Git for version control and gained practical experience with software development practices in a fast-paced startup environment.
Selected Teaching and Mentoring
Lecturer | Math School for Gifted Students | 2022 – Present
- Regularly deliver lectures and problem-solving sessions on advanced mathematics to prepare high-school and middle-school students for competitions; also serve as a lecturer during math camps held twice a year.
- Organized by The Association of Mathematicians of Sarajevo Canton.
Tutor | MetaMath | 2022 – Present
- Tutoring on advanced mathematical concepts and competition problem-solving techniques for the MetaMath training program (supported by Croatian Mathematical Society).
Selected Awards from Math Competitions
- Mediterranean Mathematics Competition (Peter O’Halloran Memorial): Bronze (2021)
- Federation of Bosnia & Herzegovina: Ranked 2nd (2021)
- Mediterranean Mathematics Competition: Honorable Mention (2019)
- Federation of Bosnia & Herzegovina: Ranked 1st (2019)
- Mediterranean Mathematics Competition: Bronze (2018)
- Federation of Bosnia & Herzegovina: Ranked 3rd (2018)
Personal Projects
Solving Markov Decision Processes Using Linear Programming:
- Implemented a solution to a toy problem using linear programming based on Bellman optimality equations.
- (Link to relevant blog post here once written)
Solving the Cutting Stock Problem:
- Compared performance of different algorithms for solving the 1D cutting-stock problem with focus on column generation approach.
- (Link to relevant blog post here once written)
Skills
- Key Strengths: Research, teaching, creative problem solving, technical communication, analytical thinking
- Programming: Python, C++, Git, LaTeX
- Languages: Bosnian (native, also Croatian and Serbian), English (C1)