Panagiotis Mokos — Portfolio

About Me

Applied Informatics graduate (ΠΑΜΑΚ, 2026) in Thessaloniki, Greece with a strong focus on Machine Learning, Data Science, and Python backend development. Experienced in classical ML — supervised and unsupervised learning, model selection, and hyperparameter optimisation — as well as deep learning pipelines. Also experienced with SQL/relational databases and frontend technologies including React and Next.js. Currently seeking an MSc in Machine Learning / Artificial Intelligence.

Experience

Education

Projects

Hyper-TA

Technical Analysis Signal Generator & Hyperparameter Optimization Framework/Library.

AbstractML

A small, abstracted ML toolkit to make it easy to perform ML tasks — data, models, evaluation, tuning, and pipelines in one unified interface.

Bitcoin Mathematical Modeling

Price prediction using Machine Learning & Mathematical Models.

Neural Networks for Deepfake Detection

Identifying manipulated facial images using CNNs & Transfer Learning.

Decentralized AI — University Thesis

Bittensor (τ) thesis: The Evolution of Intelligence / Capitalism. Simulating decentralized AI subnet mining and validator scoring offline.

Fuel Station Website

Fully responsive full-stack website for business operations.

Finpipe

Python API library that collects and aggregates market data from multiple fragmented providers into a single unified interface.

Unipmx

Unified prediction markets Python API library providing a single interface across Polymarket, Kalshi, Limitless, and other venues.

YTmation

YouTube automation framework for repetitive video tasks (cut, edit, caption). AI-agent-first — plug in your own models to generate videos and animations on the fly.

Research

Mining Intelligence (BSc Thesis)

A Bittensor thesis on decentralized AI — simulating subnet mining, validator scoring, and the evolution of intelligence offline.

Hyper-TA Official Paper

Hyper-TA: A Framework for Hyperparameter Optimization of Technical Analysis Signal Generators — towards systematic, data-driven trading strategy discovery.

BTC Price Prediction — ML Term Project

Term project for Machine Learning (7th semester): predicting future Bitcoin prices using regression. University of Macedonia, 2025.

Binary Face Classification — Neural Networks Term Project

Term project for Neural Networks: binary face classification with deep learning. University of Macedonia, 2025.

Neural Networks Term Project — Binary Face Classification

Skills

Backend

Python, FastAPI, SQL (PostgreSQL), REST APIs.

ML / Data Science

Pandas, NumPy, Scikit-learn, PyTorch, Optuna, Matplotlib, Plotly.

Web

React, Next.js, TypeScript, Tailwind, HTML/CSS.

Finance / Quant

TA-Lib, finta, yfinance.

Tools & Environment

Git, GitHub, Jupyter, VS Code, Linux, ffmpeg, ComfyUI.