George M.

I build production AI systems — working systems that ship to real clients and solve real problems, not prototypes that sit in a sandbox. I also work with founders and growing companies as a Fractional CTO, owning the full technical function, making architecture and hiring decisions, and translating business goals into a technical roadmap that can actually be executed. If you need a senior technical partner rather than just someone to build a defined spec, that is a role I am comfortable in. Currently Chief Data Officer at Nexus Lumina in Johannesburg, leading a team of eight data scientists and ML engineers across financial services, healthcare, and government clients. I operate through a UAE-registered entity (Telios Holdings FZE) — contracts, invoicing, and payments are clean regardless of your location. AI agents and LLM systems: Production agentic systems using Anthropic Claude, OpenAI, LangChain, and LangGraph — multi-agent architectures with tool use, memory, and structured output. RAG pipelines for document intelligence and knowledge-grounded generation. At Nexus Lumina I deployed a RAG system for a banking client that cut support escalations by 20%. At Liminal Clarity the LLM fraud detection platform I built from scratch attracted VC investment and scaled to multiple institutional clients. Computer vision and eKYC: National-scale identity verification platforms with document OCR (Regula SDK), face recognition, and liveness detection. Automated loan processing systems handling thousands of applications per minute. Visual inspection systems for industrial use cases. Machine learning engineering: Full lifecycle in Python and R — scikit-learn, XGBoost, LightGBM, TensorFlow, Keras, PyTorch, SparkML. Four years at inQuba building ML for health insurance and financial services: 5-class journey classifier at 95%+ accuracy, survival analysis models for churn prediction, NLP pipelines delivering 20% improvement in customer satisfaction scores, A/B tested segmentation models. SHAP and LIME explainability for every regulated or executive-facing deployment. Data engineering: Data lakes on Azure Databricks, GDPR-compliant ETL with Great Expectations, Terraform-managed infrastructure across AWS, Azure, and GCP. Built data engineering and data science teams from scratch at IT Master, reducing deployment time 40%. Implemented GDPR-compliant encryption on AWS Athena for Raiffeisen Bank across European multi-market operations. Full-stack AI products. FastAPI back end, React and Next.js front end, Supabase, Docker, Terraform, AWS Lambda and Step Functions. I build the whole thing when the engagement needs it. Active independent projects: Body of Work (multi-agent health coaching platform — Medical Advisor, Strength Coach, Nutritionist, Recovery Specialist, VO₂ Coach, LangGraph orchestration, Apple Watch biometrics); Fairways and Frontiers (luxury golf travel platform, Next.js/WebGL, AI content agents, email automation, SEO, analytics — commercially activated at a LIV Golf event); RAG Experimentation Lab (Milvus, Neo4j, Redis, Ollama, Docker on M1 Mac Mini). Stack: Python, R, SQL, TypeScript — TensorFlow, Keras, PyTorch, scikit-learn, XGBoost, SparkML — LangChain, LangGraph, OpenAI, Anthropic, Bedrock, HuggingFace — AWS (SageMaker, Lambda, Athena, S3), Azure ML, Databricks, GCP — Docker, Terraform, Airflow, CI/CD — PostgreSQL, BigQuery, PySpark, Neo4j, Milvus, Supabase — React, Next.js, FastAPI — Power BI, Tableau, Plotly. MSc Data Science (Deakin University). Postgraduate Certificate in Data Science and Business Analytics (UT Austin). AWS Certified Solutions Architect. Anthropic Claude AI Enterprise certified. Available for short-term builds, longer engagements, and fractional CTO or lead roles. I communicate clearly, deliver on time, and care about whether the thing works in production - not just whether it looked good in the demo.