Profile Avatar

Data Science & Machine Learning Portfolio

Focused on data-driven solutions, statistical analysis, machine learning models, deep learning systems, and modern web development.

View Projects Download CV

About Me

            
            class Developer:
                def __init__(self, name):
                    self.name = name
                    self.skills = [
                        "Python",
                        "Data Science",
                        "Machine Learning",
                        "Neural Networks",
                        "Problem Solving",
                        "Creative Thinking",
                    ]
                    self.mindset = "Learning by building"
                    self.mission = "Master AI concepts and create real-world intelligent solutions."
                    self.fun_fact = (
                        "Started learning C# in 2015 and quit. "
                        "Rediscovered programming with Python in 2023 and unlocked a passion for AI!"
                    )

                def current_focus(self):
                    return
                    "Deepening knowledge in ML models, neural networks, and data analysis."

                def introduction(self):
                    print(f"Hello! I'm {self.name}")
                    print("AI-driven developer passionate about data and intelligent systems.")
                    print(f"Skills: {', '.join(self.skills)}")
                    print(f"Mindset: {self.mindset}")
                    print(f"Mission: {self.mission}")
                    print(f"Currently: {self.current_focus()}")
                    print(f"Fun fact: {self.fun_fact}")

            if __name__ == "__main__":
                me = Developer("Plamen Svetoslavov")
                me.introduction()

Technical Skills

Hands-on experience spanning Python backend development, data analysis, and applied AI workflows

🧠 Python Web Development

Applied
  • Django
  • Django REST Framework
  • PostgreSQL
  • Supabase
  • MongoDB
  • Firebase
  • ORM & Query Optimization
  • Authentication & Permissions
  • Docker
  • Gunicorn
  • Nginx
  • CI/CD
  • Azure
  • JavaScript

📊 Data Science

Applied
  • Python
  • Pandas
  • NumPy
  • Data Viz
  • Statistics

🤖 Machine Learning

Applied
  • scikit-learn
  • Feature Eng.
  • Model Eval
  • Cross-Validation

🧠 Deep Learning

Applied
  • PyTorch
  • Neural Nets
  • CNNs
  • Training & Tuning
Work

Personal Projects

Data and AI projects built through hands-on experimentation and learning

goriva.online

goriva.online

Real-time fuel price tracking platform for Bulgaria. The website aggregates fuel price data from gas stations and visualizes it through tables, filters, and an interactive map, enabling users to quickly compare prices, find the cheapest stations, and explore fuel price trends across different regions and cities.

This project focuses on collecting and visualizing fuel price data from gas stations across Bulgaria.

Key Objectives:
• Collect and aggregate fuel price data from gas stations
• Provide clear visualization of prices by region and city
• Enable users to quickly identify the cheapest fuel options
• Support community submissions to keep prices up to date

  • JavaScript
  • HTML
  • CSS
  • Supabase
  • Google Analytics
  • Google AdSense
  • Role:Web Developer
  • Stack: JS · HTML · CSS
  • Year: 2026
View Repository Visit Website
Object Detection Project Preview

Object Detection with YOLO

Real-time product detection system built with the YOLO (You Only Look Once) architecture. The model predicts bounding boxes and class probabilities in a single forward pass, enabling fast and accurate multi-object detection.

This project focuses on training a custom YOLO model to detect and classify four product categories: Twix, Snickers, Bounty, and Mars.

Key Objectives:
• Train a custom YOLO model on a labeled dataset
• Achieve precise bounding box localization
• Ensure high classification accuracy across all classes
• Support inference on images and video streams

  • Python
  • YOLO
  • Role: Python Engineer
  • Stack: Python · YOLO
  • Year: 2026
View Repository
pca analysis

PCA, K-Means and Random Forest Customer Segmentation

Machine learning project for analyzing customer behavior and building segmentation models

This project focuses on analyzing credit card customer data to identify distinct customer segments using unsupervised and supervised machine learning techniques.

Key Objectives:
• data cleaning and Exploratory Data Analysis (EDA)
• Principal Component Analysis (PCA) for dimensionality reduction
• K-Means clustering to identify customer groups
• Random Forest to evaluate which original features most strongly differentiate the clusters

  • Python
  • Pandas
  • NumPy
  • Scikit-learn
  • Matplotlib
  • Seaborn
  • Role: ML Developer
  • Stack: Python · ML · Data Analysis
  • Year: 2026
View Repository

Comparative Analysis of Heart Disease Risk Factors and ML Model Performance

This project focuses on the analysis and comparison of two independent datasets related to heart disease, with two main objectives:

Identifying and evaluating key risk factors associated with the disease, and training and comparing predictive models (Logistic Regression and Random Forest) on both datasets. The workflow includes data preprocessing, exploratory data analysis (EDA), statistical hypothesis testing, and machine learning modeling, followed by a comparative evaluation of both datasets and model performances.

  • Python
  • Pandas
  • scikit‑learn
  • EDA
  • Explainability
  • Role: Data Scientist
  • Stack: Stats · ML · Viz
  • Year: 2025
Repository

Diabetes Prediction Using Health Indicators

This project focuses on building a machine learning model that predicts the likelihood of diabetes based on various health indicators.

The goal is to analyze, preprocess, and compare different feature sets to understand the most important factors influencing diabetes risk.

  • Python
  • Pandas
  • Random Forest
  • Linear SVC
  • Classification
  • Role: ML Engineer
  • Year: 2025
Repository

GTA Manager

Django-based internal management platform developed as an educational project for learning and practicing full-stack web development.

The platform is not deployed and was built solely for educational purposes.

  • Python
  • Django
  • Role: Python Web Developer
  • Stack: Python · Django · PostgreSQL
  • Year: 2024
Repository

K-means Clustering

A minimal and clear implementation of the K-Means clustering algorithm.

The project demonstrates how to group data points into K clusters based on similarity, using iterative centroid updates until convergence. Includes data preprocessing, visualization of clusters, and configurable parameters such as number of clusters and initialization method.

  • Pandas
  • Numpy
  • Matplotlib
  • Sklearn
  • Year: 2025
Repository

E-Stay Gen

Real Usage - XML Generator for the Bulgarian NRA (Electronic Transport Declarations for Stay - ETDS)

E_STAY_GEN is a desktop Python application with a graphical interface that automatically generates XML files in compliance with the requirements of the Bulgarian National Revenue Agency (NRA) for stationary transport vehicles carrying fuel. The application extracts data from an input XML file (ETD), supplements it with address details provided by the user, and creates a new stayTransportDeclaration XML file.

  • Python
  • Tkinter
  • XML.etree
  • Role:Python Engineer
  • Stack: Python · Tkinter
  • Year: 2025
Repository

🟢 Aurora Speech Extractor

Version: 2.0

Author: Pynexor Ltd.

License: MIT License

✅ What is Aurora Speech Extractor? Aurora Speech Extractor is a modern, feature-rich desktop application that converts speech from audio and video files into written text using advanced AI technology. Built with a beautiful, intuitive interface, it supports multiple output formats, real-time progress tracking, and drag-and-drop functionality.

  • Python
  • Whisper
  • Tkinter
  • moviepy
  • Role:AI Engineer
  • Year: 2025
Repository

Get in touch

Open for collaboration, freelance work, or full-time opportunities.

Let’s talk

The fastest way to reach me is via email or LinkedIn. I usually respond within 24 hours.

Email me LinkedIn

Direct contacts