Data Scientist | AI Engineer | Generative AI | LLM Applications | Time-Series Forecasting

I build practical AI systems across machine learning, forecasting, NLP, retrieval, RAG, and production API workflows. My work spans model development, evaluation, data pipelines, LLM applications, and full-stack tools that make AI systems easier to use in real projects.

• Based in Thessaloniki, Greece
• MSc in Data and Web Science, Aristotle University of Thessaloniki
• 4+ years of experience in data science, forecasting, ML, NLP, and production AI systems
• Currently focused on LLM applications, RAG, semantic search, agentic workflows, and AI engineering

• Production RAG systems for grounded answers over private or domain-specific knowledge
• LLM-powered assistants for document processing, retrieval, decision support, and workflow automation
• Forecasting and predictive analytics pipelines for retail, supply chain, and business planning
• Custom ML/NLP infrastructure including embeddings, tokenizers, vector search, rerankers, APIs, and evaluation tools
• Full-stack AI applications using Next.js, TypeScript, Python, FastAPI, PostgreSQL, Prisma, and cloud deployment workflows

A secure AI second-brain system for document intelligence, cited RAG, agentic workflows, MCP integrations, reusable skills, and LLM observability across local and cloud models.

Tech: Python, LLMs, RAG, agents, MCP, retrieval, observability

End-to-end local retrieval pipeline combining BM25 sparse search, FAISS dense search, hybrid ranking, a custom scikit-learn reranker, evaluation metrics, CLI tooling, and a Streamlit demo.

Tech: Python, BM25, FAISS, hybrid search, reranking, scikit-learn, Streamlit

A lightweight educational vector database built from scratch with dense vector storage, metadata filtering, exact similarity search, graph-based approximate search, JSON persistence, examples, tests, and benchmarks.

Tech: Python, vector search, ANN search, metadata filtering, algorithms

Domain-adaptive Sentence-BERT fine-tuning for semantic search and practical matching workflows such as product search, support tickets, job/CV matching, and retrieval evaluation.

Tech: Python, Sentence-BERT, embeddings, semantic search, fine-tuning, NLP

From-scratch tokenizer engineering project implementing BPE, byte-level BPE, WordPiece, and SentencePiece-style unigram tokenization with benchmarking, FastAPI serving, Streamlit visualization, and CI-tested package design.

Tech: Python, NLP, tokenization, BPE, WordPiece, SentencePiece, Hugging Face, FastAPI

AI-powered document conversion tool for turning PDFs into clean Markdown for documentation, knowledge-base, and RAG preparation workflows.

Tech: TypeScript, PDF processing, Markdown, AI tooling, document processing, RAG preparation

Full-stack AI/Data Science portfolio with a database-backed project showcase, protected admin dashboard, GitHub metadata sync, contact management, custom analytics, and Vercel/Neon deployment.

Tech: Next.js, TypeScript, React, Tailwind CSS, Prisma, PostgreSQL, Auth.js, Resend, Vercel

Full-stack Next.js game with PostgreSQL-backed questions, voting, mode selection, Prisma migrations, import/seed tooling, duplicate prevention, and a deployed web experience.

Tech: Next.js, TypeScript, PostgreSQL, Prisma, Vercel, Neon

Master's thesis project focused on generating plain-language summaries of clinical trial outcomes to make medical information more accessible.

Tech: NLP, transformers, DistilBERT, Mistral-7B, summarization, evaluation

Demand forecasting challenge solution for predicting daily probabilistic demand distributions across item-store combinations.

Tech: time series, forecasting, machine learning, probabilistic modeling, retail demand

Programming and data: Python, SQL, TypeScript, Java, C++, Pandas, NumPy, Jupyter

Machine learning and forecasting: scikit-learn, XGBoost, LightGBM, statistical forecasting, time-series analysis, feature engineering, backtesting, model evaluation

Deep learning and NLP: PyTorch, TensorFlow, Hugging Face Transformers, Sentence-BERT, embeddings, semantic search, summarization, information extraction

Generative AI and retrieval: OpenAI API, LangChain, MCP, RAG, vector databases, FAISS, prompt engineering, LLM evaluation, agentic workflows

Backend, MLOps, and cloud: FastAPI, PostgreSQL, MongoDB, Prisma, Docker, Kubernetes, AWS, CI/CD, Vercel, Neon

Frontend and product engineering: Next.js, React, TypeScript, Tailwind CSS, full-stack app architecture, admin dashboards, analytics

Data Scientist, Deus Ex Machina | Sep 2025 - Present

• Building AI applications with OpenAI APIs, LangChain, FastAPI, and cloud-native architectures
• Developing RAG systems with embeddings, vector databases, semantic search, and retrieval evaluation
• Creating agentic workflows and AI assistants for information retrieval, document processing, and decision support

Data Scientist, Dataviva | Nov 2022 - Oct 2024

• Built forecasting and predictive analytics solutions for retail and supply chain datasets
• Developed ensemble ML models, automated pipelines, backtesting workflows, and reporting systems
• Led proof-of-concept initiatives and technical demonstrations for enterprise clients

Application Consultant / Engineer, Veltio | Oct 2021 - Nov 2022

• Developed client-specific analytics and forecasting applications for supply chain optimization
• Translated operational requirements into data-driven solutions and supported implementation in customer environments

• MSc Data and Web Science, Aristotle University of Thessaloniki
• BSc Computer Science, Aristotle University of Thessaloniki
• Erasmus+ Computer Science / Software Engineering, Malardalen University

• Greek: native
• English: C2 proficiency

• Building production-ready LLM and RAG applications
• Improving retrieval quality with hybrid search, reranking, embeddings, and evaluation
• Developing AI assistants for document intelligence and workflow automation
• Sharing more practical ML, NLP, and Generative AI projects on GitHub

Thanks for visiting.