Ernani Hazbolatow
📊 Exitpolling

Ernani Hazbolatow /quant, data scientist, pollster/

Hi, welcome to my landing page!

I am a first-year Computational Science graduate student at the University of Amsterdam and the Free University of Amsterdam. I intend to concentrate mostly on the use of computational techniques in the financial markets during my studies. In addition, I hope to gain a rich understanding of stochastic calculus and high performance computing. Since this is a two-year program, my anticipated graduation year is 2026.

Next to my studies, I work part-time as a data scientist for Ipsos I&O Netherlands. Previously, I worked full-time as a data scientist. At Ipsos I&O, I am primarily responsible for the development of new data science methods and the optimization of in-house data science methods. I am one of few pollsters responsible for polling activities at Ipsos I&O. Next to co-organizing the polls, I investigate possible improvements to current modelling strategies.


Prior, I studied psychology with a concentration in psychological methodology and data science at Tilburg University. For my minor, I succesfully completed hand-picked courses on neural networks, mathematics, programming and behavioral economics. I graduated Cum Laude with a GPA of 8.23 (Top >5%) and 186 out of 180 ECTS.


📍 Projects

Improving L-BFGS with Nesterov’s Momentum in R
[Still developing 🟨]
I am currently developing an enhanced Limited-memory Broyden–Fletcher–Goldfarb–Shanno (L-BFGS) optimization algorithm by incorporating Nesterov's Accelerated Gradient (NAG) momentum, implemented in R. By combining L-BFGS's efficiency with Nesterov's momentum for faster convergence, the hybrid algorithm aims to improve performance in high-dimensional optimization tasks. By modifying the current state-of-the-art C++ code and porting this code through Rcpp, I expect to see moderate performance gains while retaining the stability of L-BFGS. ... See More
Exit Poll for the 2023 Dutch General Election on behalf of the Dutch Broadcasting Network (NOS) and RTL Nieuws.
[Finished 🟩]
As a data scientist at Ipsos Netherlands, I contributed to the Exit Poll for the 2023 Dutch General elections in collaboration with a senior data scientist. The combination of strategic sample selection, historical data modeling, stakeholder collaboration, and technical enhancements culminated in a robust and reliable exit poll model for the 2023 Dutch General elections, despite the challenges posed by the historical data being from 2021, a period marked by the COVID-19 pandemic. ... See More