Machine Learning Practitioner & Systems Architect
Building Autonomous Agentic Systems, Scientific Machine Learning & Safety-Critical Architectures
Brussels, Belgium & Nicosia, Cyprus

I am a computer scientist and machine learning specialist working at the intersection of autonomous agentic architectures, scientific deep learning, and safety-critical distributed systems.

Holding an M.S. in Computer Science (Computational Perception and Machine Learning) from the Georgia Institute of Technology and a B.S. in Electrical and Computer Engineering from Hofstra University, my work bridges rigorous mathematical foundations with production software craftsmanship. Alongside software development, my background includes 16+ years working within complex European safety-critical systems and technical regulatory oversight—instilling an engineering philosophy centered on algorithmic precision, test-driven validation, and verifiable system resilience.

• Autonomous Agents & Tool Ecosystems: Composable agent frameworks, adaptive execution gates, dynamic multi-tier model routing (fast, smart, reasoning), and sandboxed tool execution via the Model Context Protocol (MCP).
• Frontier LLM Orchestration & Evaluation Benchmarks: Vendor-neutral multi-provider pipelines, automated quantitative evaluation gates (0–100 scorecards), semantic intent routing, and sliding-window rate-limiting guardrails.
• Physics-Informed Machine Learning (PINNs): Integrating physical principles (aerodynamics, atmospheric thermodynamics) into deep learning loss functions and gradient-boosted trees for high-stakes modeling.
• Safety-Critical Distributed Networks: Peer-to-peer event coordination, distributed slot allocation algorithms, and immutable audit telemetry with zero single points of failure.
• Regulatory Knowledge Retrieval & Policy Analytics: Domain-specific RAG indexing dense statutory corpora (e.g., EUR-Lex) and interactive simulation platforms modeling the economic dynamics of European environmental policies.

• AeroDeck — Agentic AI & Autonomous Systems Framework
  A domain-agnostic agent execution framework built for Google Antigravity. Features an Adaptive Execution Gate dynamically switching between rapid single-turn execution and peer-reviewed multi-subagent pipelines, a resilient multi-provider router integrated with the Hermes model catalog, and custom MCP servers for browser automation and workspace tooling. Validated with a 5-tier QA pyramid including 31 RFC 6455 WebSocket framing tests and 21 security audit suites.
  TypeScript Node.js Model Context Protocol RFC 6455 WebSockets Jest

• PressHub — Enterprise AI Co-Pilot & Multi-Provider LLM Engine
  An enterprise AI orchestration engine integrating frontier foundation models with multimodal generation. Features an automated 0–100 readiness evaluation gate quantitatively scoring drafts against structured rubrics before promotion, a unified Provider Store with strict precedence resolution, semantic classification routing, and per-user sliding-window rate-limiting guardrails.
  PHP 8.2+ TypeScript Multi-Provider LLM APIs PHPUnit (73 suites) Jest WP-Cron

• logosAI — Modern Greek Neural Speech Synthesis
  Deep learning text-to-speech (TTS) engine and processing pipeline engineered for natural Greek language phonetics and expressive prosody.
  TypeScript Neural Audio Synthesis NLP

• PRCXGBoost — Physics-Informed Neural Networks & Flight Telemetry ML
  Scientific machine learning pipeline combining Physics-Informed Neural Networks (PINN in PyTorch Lightning) with XGBoost, METAR atmospheric parsing, and BADA aerodynamics for aircraft fuel consumption prediction (^2 = 0.961$, RMSE = 204.6 kg). Peer-reviewed and published in the Journal of Open Aviation Science (2026).
  PyTorch Lightning PINNs XGBoost Aviation Telemetry Scikit-Learn

• ddfn_v2 — Decentralized Air Traffic Flow Management Network
  A decentralized, peer-to-peer drop-in alternative for centralized air traffic flow management. Utilizes a distributed pub/sub SWIM network to coordinate traffic flows, compute CASA delay allocations, and renegotiate 4D flight trajectories with immutable SQLite audit logging.
  Python FastAPI WebSockets Pydantic v2 Distributed Systems

• AIXM-Parser — Aeronautical Information Exchange Model Parser & Visualizer
  High-performance Python library and interactive browser tool for parsing AIXM 4.5 XML datasets, extracting complex airspace geometry, and rendering interactive aeronautical map overlays.
  Python GIS / Geospatial XML Parsing Aviation Data

• AirConnectivity — Graph-Theoretic Air Transport Network Topology
  Network graph analytics computing air connectivity indicators using PageRank algorithms and onwards network connectivity metrics across flight network topologies.
  Python NetworkX Graph Algorithms Data Science

• llama-eurlex — EU Law Legal RAG Assistant
  Domain-specific legal tech assistant indexing European Union legislative corpora (EUR-Lex). Integrates live SOAP/WSDL webservices with LlamaIndex and local HuggingFace transformers for automated query-driven statutory cross-referencing and compliance verification.
  Python LlamaIndex EUR-Lex SOAP/WSDL Transformers Vector Stores

• FuelCost — EU Decarbonization Policy Simulation Platform
  Interactive analytics platform modeling the multi-variable economic and operational impacts of European environmental policy packages (Fit for 55, SAF blending mandates, ETD fuel taxation, and EU ETS carbon compliance mechanisms).
  Python Dash / Plotly Policy Analytics Regulatory Modeling

ext Languages: Python, TypeScript, PHP, SQL, Shell, C/C++ Machine Learning: PyTorch Lightning, XGBoost, Scikit-Learn, LlamaIndex, HuggingFace Transformers Agentic & LLM: Model Context Protocol (MCP), Multi-Provider API Orchestration, Evaluation Rubrics Distributed & Web: FastAPI, Node.js, WebSockets (RFC 6455), SQLite, Pydantic v2, REST / SOAP Engineering: Test-Driven Development (TDD), Docker, GitHub Actions CI/CD, Linux

• M.S. in Computer Science (Computational Perception and Machine Learning Specialization) — Georgia Institute of Technology
• B.S. in Electrical and Computer Engineering — Hofstra University
• 16+ Years in Complex Systems & Safety Governance: Extensive background bridging advanced software engineering with international and European safety-critical systems, technical standards, and institutional regulatory oversight across aviation (Eurocontrol, EASA) and aerospace security (Galileo SAB).

• LinkedIn: linkedin.com/in/eftychios-eftychiou-88686843
• Email: eftychios.eftychiou@gmail.com
• Location: Brussels, Belgium & Nicosia, Cyprus