Staines, UK / Royal Holloway AI MSc candidate

Dimitrios Devliotis

Software developer building practical AI systems, games, clean web products, security tooling, and interactive experiences with a strong design mindset.

Read case studies See projects
Current focus RAG systems, game development, computer vision, and portfolio-grade engineering.
Game development Unity, C#, procedural generation, enemy AI behaviour, UI, and performance optimisation.
Strengths Problem solving, ownership, creative UI thinking, and resilient teamwork.

Selected work

Projects with clear technical signals for employers.

A curated view of recent GitHub work, academic projects, and practical tools. Live repository dates are hydrated from GitHub when available, with game-development and AI work brought forward for employers.

Case studies

How I approach problems, not just what I built.

These are the strongest employer signals on the portfolio: applied AI, computer vision, simulation, testing, and clear technical decision-making.

Applied AI

RAG-Based FightIQ System

A Python retrieval augmented generation workflow integrating web search, document ingestion, and LLM-based response generation for domain-specific questions.

Problem

Domain-specific questions need current, relevant context. The goal was to reduce ungrounded answers by retrieving useful source material before generating a response.

Approach

  • Combined document ingestion, web scraping, text extraction, and structured storage for dynamic knowledge retrieval.
  • Designed the response flow around retrieval and prompt engineering to improve answer grounding.
  • Used the project to deepen understanding of end-to-end AI systems and hallucination mitigation.

Employer signal

Shows practical AI engineering: data preparation, retrieval design, model integration, local-first thinking, and readable project structure.

Python NLP RAG Web scraping LLM integration View repository
Game development

Asteroid Miner

A Unity/C# 3D shooter built around procedural chunk generation, real-time enemy AI behaviour, object pooling, spawning systems, and responsive UI.

Problem

Real-time games need systems that feel responsive while staying efficient. The project focused on building a playable loop while managing generation, enemies, UI, and performance constraints.

Approach

  • Built procedural chunks to create a dynamic play space rather than a fixed map.
  • Implemented enemy spawning and AI behaviour to create pressure and combat flow.
  • Used object pooling and structured gameplay systems to improve runtime performance.

Employer signal

Shows Unity/C#, gameplay architecture, systems thinking, optimisation, and the practical skills needed for junior game-development roles.

Unity C# Game AI Procedural generation Object pooling
Computer vision

LiftLens

A fitness form analysis prototype that uses pose landmarks and geometry to inspect movement from video.

Problem

Exercise feedback is useful only when it is easy to capture. LiftLens explores how ordinary video can become structured movement feedback without specialist hardware.

Approach

  • Used MediaPipe pose landmarks and OpenCV frame processing to detect body positions.
  • Calculated joint angles with vector geometry to reason about movement quality.
  • Kept the prototype focused on a squat sample so the core tracking loop could be validated first.

Employer signal

Demonstrates computer vision, mathematical reasoning, prototyping discipline, and the ability to turn AI output into user-facing feedback.

Python MediaPipe OpenCV View repository
Machine learning

World Cup Predictor

A Python tournament simulator that estimates 2026 World Cup win probabilities through generated match data, ML classification, and Monte Carlo runs.

Problem

Tournament outcomes are uncertain and depend on both team strength and bracket paths. The project needed a repeatable way to simulate that uncertainty.

Approach

  • Generated synthetic match data with a Poisson goal model calibrated around international football assumptions.
  • Trained a logistic regression model using ELO-based features and knockout context.
  • Ran full tournament simulations many times and backed the code with pytest unit tests.

Employer signal

Shows modelling judgment, simulation thinking, CLI structure, testing habits, and clear explanation of probabilistic results.

Python scikit-learn Monte Carlo pytest View repository

GitHub pulse

Recent public activity, kept fresh.

Loading current repositories from GitHub...

Capabilities

Engineering range with an AI direction.

The stack reflects both academic foundations and self-directed builds: from Python ML tooling to React interfaces, C# gameplay systems, backend services, databases, and cloud-aware development.

Languages

Python, C#, JavaScript, TypeScript, C++, Java

Comfortable moving between scripting, frontend, object-oriented systems, and application logic.

Game development

Unity, C#, AI behaviour, procedural systems

Interested in game development roles and experienced with mechanics, enemy systems, UI, optimisation, and real-time architecture.

Web and product

React, Angular, Next.js, APIs, responsive UI

Focused on clear flows, fast pages, recruiter-friendly storytelling, and accessible interfaces.

Back-end and systems

Node.js, Express, .NET, SQL, MongoDB, AWS

Developing stronger full-stack foundations across APIs, databases, networking, systems architecture, and cloud deployment concepts.

Story

A practical path into software and AI.

MSc Artificial Intelligence

Royal Holloway, University of London. Current focus includes AI systems, optimisation, and applied ML projects.

BSc Computer Science

Royal Holloway, University of London. Covered software engineering, algorithms, databases, operating systems, cryptography, and network security.

Computer Science Society member

Royal Holloway. Engaged in technical discussion, peer learning, AI/software topics, and collaborative sessions.

Benugo, Egham

Built strong communication, multitasking, ownership, and pressure-handling skills in a fast-paced service environment.

Web Security seminar speaker

Presented web security principles at a Computer Science Student Convention in Greece.

CV

Current CV and full background.

The CV includes education, selected projects, game-development experience, full-stack skills, work experience, language fluency in English and Greek, and additional eligibility details.

Open resume PDF

Contact

Ready for software, AI, and graduate developer opportunities.

I am especially interested in roles where I can combine problem solving, practical engineering, game-development thinking, AI systems, and user-focused product work.

Email dimitrios.devliotis@gmail.com LinkedIn linkedin.com/in/dimitriosdevliotis GitHub github.com/Azzurre