Matthaios Markatis
Data Scientist | ML Engineer | Systems Developer
✨About Me
BSc Physics graduate from the University of Sheffield with strong expertise in computational methods including numerical simulations and algorithm optimization, data analysis using advanced statistical techniques, and full-stack software development. Certified in IBM Data Science and AI Engineering, with practical achievements such as deploying a traffic light identification system with 98% accuracy. Demonstrated success in developing impactful, end-to-end solutions including an IoT-enabled wildfire detection system achieving 86% predictive accuracy and a mobile application featuring fine-tuned Diffusion models for personalized image generation. Proficient in building robust data pipelines, deploying optimized predictive models, and effectively integrating AI into scalable software solutions using cloud platforms (AWS, Azure) and advanced DevOps methodologies (CI/CD, Docker, Kubernetes). Passionate about driving technological innovation through strategic applications of AI, Data Science, and comprehensive systems engineering.
⚡Technical Skills
Programming & Development
Machine Learning & AI
Data Science & Engineering
Web, Cloud & Tools
Hardware & Systems
🎓Education & Certifications
BSc Physics
University of Sheffield
2:1 Classification (Upper Second-Class Honours)
Programming & Technical Modules
Physics Core Modules
IBM AI Engineering Professional Certificate
Coursera
Comprehensive program covering machine learning, deep learning, and AI engineering with hands-on projects. Learned techniques in RL, Hyperparameter tuning, Model selection and architecture along with deployment pipelines such as AWS SageMaker.
Key Achievement:
Deployed a traffic light identification system with 98% accuracy using IBM Watson and TensorFlow.
IBM Data Science Professional Certificate
Coursera
Intensive data science program covering the complete workflow from data collection to insights delivery. Learned techniques in Data Analysis, Data scraping and cleaning and full data to model development strategies along with ETL pipelines using AWS, Azure and Github Actions.
Key Achievement:
Developed a SpaceX Falcon landing prediction classifier achieving 80% accuracy with a live website map through statistical analysis and feature engineering.
🚀Projects
Wildfire Detection System
University of Sheffield Research Project
Developed a real-time fire detection system built from a Raspberry Pi equipped with sensors and LoRaWAN, together with ML models (Random Forest, Decision Trees) to predict fire ignition before it occurs. Built a responsive web dashboard with Flask, SQLite, and Leaflet.js for interactive mapping. Trained on NASA FIRMS data, achieving 86% detection accuracy and 10-minute early warning capability. Deployed at the cheap cost of £46 per unit.
AI Personal Image Generation App
Flutter Mobile Application
Created and launched a production-ready Flutter app for personalized AI image generation. Implemented full-stack solution with custom fine-tuning pipeline for FLUX.1 diffusion models via Replicate API. Integrated secure user authentication (OAuth), database management (Firebase), and payment processing (Google Billing). Published to Google Play Store with ongoing user adoption.
Job Application Automator
Web Automation Project
Engineered an intelligent automation system using Python, Playwright, LangChain and Browser-Use that handles any job application process at any website, dynamically scraping and filling forms with the assistance of an LLM agent. Integrated Azure OpenAI API for dynamic cover letter generation, implemented pattern recognition for form detection and playwright for job description scraping, and built robust error handling for seamless operation. System capable of submitting 400+ targeted applications daily.
Fine-Tuned LLM Chat Bot
AI Integration Project
Built a fine-tuned LLM chatbot that was fine-tuned on Discord user messages using Azure ML, and Azure OpenAI. Developed a context-aware prompting system instilled with tools and an advanced audio processing pipeline with Coqui TTS for voice capabilities. Deployed via AWS SageMaker with a live CI/CD pipeline. Gained adoption across 20+ servers after publication to a bot marketplace.
Autonomous FPV Drone with Navigation Software
Personal Project
Designed and built a fully autonomous drone with software for GPS waypoint navigation and PID stabilization. Implemented sensor fusion algorithms together with INaV for reliable positioning and custom Image Recognition (CVS) for vision-guided objectives. Achieved autonomous flight capabilities with precise navigation within a 1-2 km range and automated return-to-base functionality.
💼Work Experience
AI Data Annotator & Reasoning Specialist
Outlier AI
Spearheading data annotation and synthetic task creation for training large language models (LLMs) using Reinforcement Learning from Human Feedback (RLHF). Specialize in annotating and engineering tasks designed to teach models graduate-level mathematics and logical reasoning. Collaborate on fine-tuning initiatives for state-of-the-art reasoning agents with focus on structured task generation, error identification, and model behavior analysis in multi-hop reasoning environments.
Team Leader
Meltdown-Wetherspoons
Led operations and staff training in high-volume venues. Optimized workflow processes, improving operational efficiency by 15% and consistently scoring 90+ % in CQSMA. Developed crisis management strategies and provided performance coaching to team members in fast-paced environments.
NHS Shadowing Experience
Sheffield Teaching Hospitals NHS Trust
Observed healthcare technology systems and electronic patient records management across various departments. Gained insight into medical data integration, diagnostic imaging workflows, and cross-departmental information systems.