Georgios B.

CONTEXT operates a data categorisation platform processing billion-row datasets across ~250 product categories and multiple global pipelines.

• Led the architecture and delivery of a complete redesign of the QC framework, integrating an AI agent with MCP tools and reducing monthly processing time by 40% (from 50 to 30 hours)
• Architected and delivered a FastAPI-based internal platform supporting RAG and machine learning inference, replacing legacy manual workflows and increasing data throughput by up to 7x
• Owned the full machine learning lifecycle, including data preprocessing, feature engineering, model training, evaluation, and deployment, while enforcing a 98% precision threshold that increased automated processing volume by 13%
• Established an ML validation framework and release management process using MLflow, DVC, Docker, and GitHub Actions, enabling safe and repeatable deployments
• Introduced code review practices, coding standards, and documentation conventions, reducing onboarding time and limiting technical debt
• Advised leadership on model selection, data quality, and LLM adoption strategy across the platform roadmap

Sunlight Group is an industrial manufacturer of energy storage and battery systems operating across 100+ countries with €266M revenue.

• Delivered a real-time part identification system into production, serving engineering drawings at assembly-line stations and optimizing it for low-power edge devices
• Engineered a drone object-tracking system that enabled autonomous flight by identifying and tracking vehicles and bicycles with dynamic motion synchronization
• Built and deployed machine learning models for object detection, linguistic analysis, and autonomous navigation using PyTorch and TensorFlow
• Optimized inference for resource-constrained Raspberry Pi devices, balancing accuracy with real-time performance on the factory floor

Systems Sunlight S.A. is an industrial energy and battery technology company developing storage and embedded power systems.

• Engineered a hybrid energy control unit using Raspberry Pi, CAN bus, and Python, integrating photovoltaic, battery, and grid power sources to optimize cost efficiency
• Designed standardized data acquisition workflows for SQL databases, CSV files, and binary telemetry data
• Evaluated AR glasses, thermal cameras, and gesture recognition hardware for industrial applications
• Supported cross-functional R&D teams in validating next-generation energy technologies and embedded systems

Democritus University of Thrace conducts NLP and document-analysis research within its Department of Electrical and Computer Engineering.

• Implemented custom neural network layers in MatConvNet, porting architectures from Caffe for word spotting and document binarisation

• Built automated dataset-generation pipelines in MATLAB to support research experiments

• Contributed to NLP and document-analysis research adopted in international benchmarks

• Developed and evaluated neural architectures for handwritten text recognition across multiple scripts

NCSR "Demokritos" is Greece's largest multidisciplinary research centre, conducting work in AI and computer vision.

• Applied deep learning and computer vision techniques to multilingual word spotting and document image binarization
• Developed and evaluated neural network architectures for handwritten text recognition across multiple writing systems
• Built evaluation benchmarks and datasets that were adopted in ICDAR and ICFHR international competitions
• Contributed to research that was later published in IEEE conference proceedings

"Athena" Research Centre is a public research institute focused on data, language, and digital technologies.

• Contributed to the EU-funded PRESIOUS project on predictive digitisation, restoration, and degradation assessment of cultural heritage objects

• Applied computer vision and 3D modelling techniques to analyse and restore historical artefacts

• Developed data-driven methods for digital conservation and predictive analytics on artefact degradation

• Supported research deliverables for an EU-funded multi-partner consortium