Persistent AI Companion
A long-running companion runtime with identity, episodic memory, self-modeling, social contingency, restart recovery, and bounded resource use.
About the researcher
Tym
Huseby
Technical product and AI systems leader building the conditions for intelligence to persist, adapt, and become more than a performance.
TYM HUSEBY / FOUNDERANIMUS MACHINAE
The work behind the question
Make the system
earn its aliveness.
I work at the intersection of product strategy, applied AI, and artificial life research.
For more than a decade, I have translated complex technical constraints into platforms people can actually use. At Animus Machinae, that same discipline turns inward: building organisms, memory systems, and environments where believable life has to come from persistent causes instead of a convincing surface.
The work is deliberately evidence-heavy. Claims stay bounded, failures stay visible, and every experiment is designed to tell us what the system is actually doing.
700+software integrations led across an enterprise payment ecosystem
100K+deployed devices supported through platform scale and enablement
80%faster partner onboarding through tooling and workflow redesign
30%measurable productivity improvement from AI-assisted operations
Selected applied AI systems
Embedded AI & Robotics
A Raspberry Pi 5 and Hailo-8 platform combining multimodal sensing, local inference, servo expression, voice interaction, and observable real-time control.
Local Knowledge Engine
A private context system combining notes, journal, calendar, budgeting, document processing, retrieval, memory, speech input, and grounded AI assistance.
Evidence-Based Governance
A deterministic engineering workflow built around explicit requirements, controlled tests, immutable audit records, and reproducible qualification criteria.
Technical toolkit
Product & platforms
Product strategy, roadmaps, requirements, platform operations, API ecosystems, developer enablement, certification, release governance, and stakeholder leadership.
AI systems
Agentic systems, RAG, vector search, contextual memory, AI evaluation, prompt and context engineering, local/private inference, QLoRA, and OpenAI-compatible APIs.
Engineering
Python, SQL, JavaScript, TypeScript, FastAPI, Next.js, Node.js, SQLite, PostgreSQL, Google Apps Script, Looker Studio, and REST APIs.
Infrastructure & edge
Linux, Docker, Git, WSL2, CUDA, Raspberry Pi 5, Hailo-8, OpenCV, sensor fusion, speech systems, and embedded behavioral control.