Engineering at NTUA, finance at Deree.

Five years of electrical and computer engineering on the AI track at NTUA. From 2024 I also took a double minor in finance at Deree, on scholarship.

Evangeliki Scholi SmyrnisEvangeliki Scholi Smyrnis
NTUA · MEngNTUA9.2 / 10Thesis
Deree · Double minorDeree4.0 / 4.0

18,960 / 20,000 Panhellenic exams · top 1%

MEng in Electrical & Computer Engineering5-year integrated Master’s · AI & Machine Learning

A lot of mathematics, probability and statistics above all, with electronics, computer systems and software built on it. On the AI track: machine learning, neural networks, image and video analysis, and forecasting.

9.2 / 10degree GPA · Top 5% of cohort
18,960 / 20,000Panhellenic exams · Top 1% of the sciences track

Machine learning that designed new antibacterial peptidesMy thesis, Sep 2025 – Oct 2026, with Houston Methodist Research Institute. 13 of the 15 made in a lab worked.Read about the research

Double Minor in Finance, and in Accounting & Financial Management

Finance and how markets work, accounting, company valuation and financial modelling, taken at the same time as engineering at NTUA.

4.0 / 4.0degree GPA

SNF & Merit Scholarships

Alongside my studies.

12,000+
Open-Access Course Notes
Downloads of the notes and solved exams I posted for 13 NTUA courses.
See the notes
2024–26
VP
Accounting & Finance Society, Deree
Vice President, after Public Relations. Events and guest speakers.
2025–26

Hackathons.

Dyania Health Hackathon 2026

Team of four · 3 days

Predicting when a replacement heart valve fails.

A team of one clinician and three engineers proposed a study to predict when a replacement aortic valve will fail, so the patients most at risk get scanned sooner.

The hospital data had only 10 failures, too few to learn from. My part was the synthetic patient cohort the models trained on, built from published trials, the scripts behind the results, parts of the study protocol, and the presentation.

517fewer echocardiograms per 1,000 patient-years than scanning everyone yearly, on synthetic data
6,000synthetic patients per cohort, drawn five times
3,580patients the study protocol needs

GitHub

Built for my courses.

TollTrax
Software Engineering
A platform that works out what 8 Greek highway operators owe each other in tolls, with a REST API, a command-line client and a React front end. Team of six; my part was the login, the API’s JSON and CSV responses, and the CLI.
- Node.js
- Express
- PostgreSQL
- React
GitHub
clearSKY
SaaS
A platform where universities publish course grades and students request reviews: 6 Node.js microservices, each with its own PostgreSQL database, linked by RabbitMQ, with a React front end. Team of six; my part was most of the front end and much of the back end.
- Node.js
- PostgreSQL
- RabbitMQ
- React
- Docker
GitHub
Deep Learning Labs
Neural Networks & Deep Learning
6 Wide ResNets trained on CIFAR-10 and tested against corrupted images with and without MixUp, then DistilBERT fine-tuned for sentiment, in PyTorch and Hugging Face.
- PyTorch
- Hugging Face
- Google Colab
GitHub
Graph Embeddings vs. GNNs
Information Systems
7 models compared on 4 graph classification datasets and written up as a short paper. Team of three; my part was the Graph2Vec and DeepWalk models.
- Python
- PyTorch Geometric
- scikit-learn
- Optuna
GitHub
Adventurize
Human-Computer Interaction
A Flutter travel app with 10 screens: photos pinned to a map, photo challenges, friends added by QR code, and a leaderboard. Team of three.
- Flutter
- Dart
- Google Maps
- SQLite
GitHub
Music Festival Database
Databases
A MariaDB database for a yearly music festival, with 22 tables, triggers for its rules, 15 analytical queries and a ticket resale procedure. Team of three.
- MariaDB
- SQL
GitHub