• Conducting experiments to evaluate the performance of multilingual language models by fine-tuning on the GreekQA dataset and transfer learning with a zero-shot cross-lingual approach on multiple datasets of other languages. • Supervising the collection of additional question-and-answer pairs for the GreekQA dataset by crowdworkers.

National & Kapodistrian University of Athens | Oct 2022 - Present

Teaching Assistant

• K08: Data Structures & Algorithms, Spring 2023 • YS02: Artificial Intelligence, Fall 2022

AI Team, National & Kapodistrian University of Athens | Jan 2022 - Sep 2022

Undergraduate Research Assistant

• Completed my Bachelor's thesis "GreekQA: A Crowdsourcing Platform and its Use for Creating a Greek Question Answering Dataset" in collaboration with the AI Team of the National & Kapodistrian University of Athens. • Computed the Greek Wikipedia’s internal PageRanks to select high-quality articles in Greek and curated passages. • Implemented a web crowdsourcing annotation platform for the collection of question and answer pairs on curated passages. • Analyzed the questions, answers, and the reasoning required to answer the questions based on the corresponding passages. • Evaluated the human performance on proposed GreekQA dataset for future experimental evaluation of language models.

METIS Cyberspace Technology S.A. | Oct 2021 - Sep 2022

Data Scientist, Machine Learning

• Undertook the continuity of the EU-funded Question Answering System in Maritime (QAMar) project. Improved and extended the custom named entity recognition and intent classification of QAMar for the purposes and functioning of a virtual assistant chatbot. • Conducted research on a vessel speed optimization project targeting fuel consumption minimization based on particle swarm optimization, time-series segmentation, and vessel performance prediction models using maritime weather forecasting. Made run time ∼ 20 times faster by delivering an implementation that uses Numba JIT compiler for Python. • Designed deep learning, tree-based ensemble, and statistical models for maritime tabular data regression and classification. • Βuilt data pipelines for the maintenance of tables with a change data capture design through jobs orchestration. • Automated the monitoring, retraining and deployment process of ML models building experimental end-to-end ML pipelines. • Implemented a vessel performance analysis and route cost prediction application for the estimated routes of vessels. • Designed and implemented interactive data visualization dashboards of maritime applications for end users.

METIS Cyberspace Technology | Jul 2021 - Oct 2021

Data Engineer Intern

• Proposed and implemented an unsupervised sensor data anomaly detection system combining statistical and deep learning techniques. Incorporated this system into a data pipeline that handles over 10 billion raw measurements per month. • Built a sensor data reception quality evaluation system that detects and assesses the lost measurements from sensor devices. • Evaluated the categorization of sensor devices by applying data profiling and multiclass classification algorithms.

National & Kapodistrian University of Athens | Feb 2020 - May 2020

Undergraduate Laboratory Teaching Assistant

• K08: Data Structures & Algorithms, Fall 2020

My Education

Take a look at my formal education

Bachelor of Science in Computer ScienceNational & Kapodistrian University of Athens | 2022

About Me

Efstathios hasn't filled in a bio text

Create Your Data Portfolio for Free