# projects

5 entries

Here, I showcase some of my major projects.

FinMLKit: A financial ML toolkit

The main goal of this python library is to provide a solid foundation for financial machine learning, enabling users to process raw trades data, generate different types of bars, intra-bar features (eg. footprints), bar-level features (indicators), and labels for supervised learning. It tackles three chronic causes of unreliable results in the field: time-based sampling bias, weak labels, and throughput constraints that make rigorous methods hard to apply at scale.

open

Digital Holographic Microscopy

Digital holography is a technique that allows the numerical reconstruction of the 3D volume from a camera image capturing the interference pattern of coherent light scattered from the sample – called the hologram. In this project, I've developed deep-learning models for object detection, segmentation, classification, focusing, phase reconstruction, and more. We have founded a startup company called HoloDetect to commercialize the technology. Besides project development, we have published several peer-reviewed papers in top-tier journals, contributing to the field of digital holography.

Impaired Speech Correction

In this proof-of-concept project, I've researched and developed speech-to-speech deep-learning models to correct impaired speech of dysarthric speakers. We've presented our results in the XVIII. Hungarian Conference on Computational Linguistics.