Athinaios
Data Scientist · AI Engineer · MSc @ UCL
Building intelligent systems from data to deployment - geospatial pipelines, multi-agent platforms, and full-stack AI applications at scale.
Projects
Geospatial pipelines, predictive models, and AI-driven analytics - built to surface insights from complex spatial data.
NeuroVault
Local-first AI memory for Claude - A Tauri desktop app that gives Claude persistent memory across conversations. Markdown vault + knowledge graph + hybrid retrieval (semantic + BM25 + graph) - outperforms RAG at 96% hit@3, ~275 tokens/answer, and runs entirely on your machine.
AEGIS
Maritime Shipping Risk Intelligence Platform - A data engineering platform aggregating maritime intelligence from 7+ sources across 4 database technologies, with LLM-powered CrewAI agents for risk assessment, route planning, and trade compliance.
London Synergy Index
Predictive Site Selection - Architected an end-to-end geospatial pipeline for 55,000 H3 hexagons, integrating LandScan satellite rasters, ONS Census data, and OSM POIs enriched with graph-centrality and competition-density features. Modeled underserved retail locations using spatially cross-validated XGBoost classifiers; applied Burt’s Structural Hole Theory to identify market gaps.
London Housing & Crime
Geospatial Analysis - Developed a Spatio-Temporal pipeline merging 1M+ records to engineer a predictive ‘Opportunity Index,’ transforming raw crime and housing data into an actionable investment strategy. Optimized an XGBoost regressor via Optuna, achieving an R² of 0.92.
RiskTerrain
S&P 500 Risk Intelligence - Engineered a 6-node LangGraph pipeline that ingests live geopolitical events (USGS, NewsAPI), performs multi-hop supply chain graph traversal across 154 companies in SurrealDB, and scores cascading risk exposure in under 8 seconds. Implemented hybrid graph-vector retrieval with sentence-transformer embeddings.
Data Engineering Pipeline
Star Schema Lakehouse - Architected a centralized Star Schema Lakehouse to unify semi-structured NoSQL and CSV datasets; utilized PyMongo and Pandas to flatten nested JSON and engineer derived geographical dimensions. Streamlined performance with columnar Parquet formats and DuckDB for high-speed SQL aggregations.
Dataportfolio.co.uk
Full Stack SaaS Platform - Launched a production-ready SaaS platform that automates portfolio creation for data professionals, featuring a ‘coding-less’ GUI to manage backend data persistence and dynamic metadata updates.
Experience & Education
Geospatial Data Scientist
Intelmatix (Dissertation Placement), London
Engineering a spatial network of 1M+ POIs using Python and NetworkX. Architecting serverless pipelines via AWS (S3, Glue, Athena) to automate processing of high-dimensional geospatial datasets at scale. Analyzing ecosystem-level insights using clustering and co-occurrence matrices.
MSc Business Analytics
UCL School of Management, London
Current Grade: 75% Distinction. Modules in Data Engineering (MLOps, Docker, AWS, Git, SQL), Operational Analytics, Machine Learning, and Predictive Analytics.
Data Scientist
Globassure, Athens, Greece
Built a data scraping pipeline to identify and extract potential client leads, securing 7 start-up businesses for employee insurance coverage. Cleaned and processed 120,000+ records, performed EDA, and surfaced actionable business insights.
BSc Business Management (Data Analytics)
Henley Business School, University of Reading
First Class Honours. Modules in Data Analytics, Machine Learning Introduction, and Information Systems. Co-Founded the Data Analytics Society.
Continuous learning
Mathematics, machine learning, data analytics - fundamentals deliberately deepened.
Mathematics for Machine Learning
Linear Algebra · Multivariate Calculus · PCA
Google Advanced Data Analytics
Statistical methods · Regression · ML modeling
Microsoft Power BI Desktop
DAX · Power Query · Data modeling
IBM Data Analyst
Python · SQL · Tableau · IBM Cognos
The whole person
Leadership, discipline, and the principles that shape how I work.
Co-Founder & President
Built the Data Analytics Society at the University of Reading from zero into a thriving 200+ member community. Hosted technical workshops, recruiter networking nights, and a hackathon partnered with industry sponsors.
Discipline & Movement
Karate Black Belt 1-dan. Active kickboxing practitioner. The same Kaizen principles I bring to dojo training - small daily improvements compounding - shape how I approach engineering: deliberate practice, candid feedback, no shortcuts.
Ready to build something extraordinary?
Geospatial intelligence, AI pipelines, multi-agent platforms - or just data science chats. Always open to interesting work.