Phil Laznicek
Summary
I design and build AI systems for places where failure has real institutional cost: hospitals, pharmaceutical work, government ministries, space systems and defence. The systems have to be auditable, sovereign and reliable in production.
I am one of the current directors of TNG. My main contribution there is technical: finding ways to apply AI so that workers spend less time on tedious tasks and more time on productive work. I also run Arytma, a small research lab in Calgary focused on sovereign and regulated AI. Before that I was the first AI architect at the Czech government’s Digital and Information Agency (DIA), at the rank of vrchní rada. Earlier I was AI and Data Architect at The Adecco Group.
Values
I work on AI that stays under human and institutional control, especially in high-stakes domains. The aim is systems that support the people making decisions rather than replace them, stay auditable, and keep critical capabilities under national or organizational sovereignty.
Core Skills & Role
- AI Architecture & Systems: Generative AI, local LLMs, RAG, agentic systems (LangGraph, AutoGen), neurosymbolic methods, Python, PyTorch, TensorFlow, on-premise and secure deployment.
- Broader Background: Solution and data architecture, full-stack development, enterprise data platforms, API design, production delivery.
- In TNG: I serve as a director. Most of my effort goes into technical solutions and into using AI to reduce repetitive work and improve productivity.
TNG & Cooperative Work
- Take part in the governance and direction of the cooperative as one of the current directors.
- Concentrate on technical solutions and on applying AI to free workers from tedious work and raise productivity.
- Work with other members on practical technology questions that come up in a worker-owned IT cooperative.
Professional Experience
Small Canadian research lab working on AI for regulated domains.
- Healthcare and pharma: Clinical NLP mapped to ICD-11, graph neural networks for pharmacovigilance, agent-based epidemiological modelling, neurosymbolic clinical trial analytics.
- Government and municipal: Secure retrieval over access-controlled documents, on-premise voice agents, agentic support for procurement.
- Space systems: AryBotanist (neuromorphic plant health monitoring for deep-space bioregenerative life support), AryHydroGuard (autonomous water disinfection for closed-loop microgravity hydroponics), AryOrbit (radiation-hardened sovereign edge AI kernel). The systems were developed in response to Canadian Space Agency requirements related to space food production programs.
- Infrastructure & Stack: Multi-agent systems with local LLMs, on-premise and air-gapped setups. Python, PyTorch, TensorFlow, LangGraph, AutoGen, vector databases, Neo4j, PostgreSQL, Azure, AWS, GCP.
- Set architectural standards for sovereign on-premise AI across state institutions, with preference for open-weight models.
- Helped move the government’s AI Competency Centre from pilots into production use.
- Designed secure RAG pipelines so ministries could query sensitive internal documents without sending data to the cloud.
- Advised on EU AI Act alignment and related risk questions.
- Global AI and data platforms for a large workforce company operating in more than 60 countries.
- Early GPT-based career assistant.
- Global Data Platform and Enterprise Data Warehouse.
- Architecture for Global Business Intelligence; dynamic contract pricing and skills mapping.
- Stack: Azure, GPT-4, Databricks, CosmosDB, MS SQL, Python, C#, Power BI.
- Executive leadership across data science, system architecture, and enterprise cybersecurity infrastructure.
- Mission-critical clearing and derivatives settlement systems during the historic European Euro monetary transition.
- Enterprise database management, data warehouse optimization, and business intelligence analytics.