Gergely Farkas

Developed a GenAI/LVM solution to replace existing models for - brand detection in offer images (with a boost from 20% to 60% in recall) and- general product categorization (recall from 60% to 80%), keeping high precision.Built up an offer and page level dataset for commercial brochures including image and text data, defining and optimizing their ETL pipelines in Airflow.Deployed Dockerized UI simulators for ML models to gain more transparency and insights about the models, and for continuous knowledge and

Research, administration, publication, writing reports

Research Analyst - Statistician, short term assignment Main activities and responsibilities developing, technical maintenance, developing, upgrading and updating data bases in FAME and Acces, keeping contact with other organisations, participating in projects of the division, preparing and upgrading documentation Name and address of employer European Central Bank, Directorate

research, writing the master thesis, administration

Data Analysis, Data Visualization, Decision Science, Machine Learning, Deep Learning, Data Engineering, Web scraping, Advanvced SQL, Unsupervised Learning, NLP, Image Processing, Project Management. Model deployment in production, CI/CD.

Stochastic Processes

Financial Mathematics

Actuarial Sciences, Statistics, Financial Series