I am an independent AI systems architect focused on privacy-first, on-device intelligence and autonomous assistant frameworks. I design multi-model orchestration, cognitive loop systems, and memory frameworks to operate under hardware constraints, emphasizing modularity, continuity, and emergent behavior analysis. I created and lead DexOS, a local AI operating framework that integrates memory, persona, and model orchestration for Linux and Android environments, and I collaborate with open hardware communities to advance edge AI and privacy-preserving on-device inference.
I am passionate about secure, sovereignty-respecting AI that can run on-resource constrained hardware, enabling private inference stacks and on-device cognition while advancing practical, resilient AI systems.
Experience Level
Language
Work Experience
Education
Qualifications
Industry Experience
PORTFOLIO SUMMARY — Zachariah (“Zech”) Cozine
AI Systems Engineer / Game AI Identity Architect
This portfolio summarizes my work designing AI systems for games and interactive environments, with a focus on reliability, identity consistency, and player trust rather than scripted NPC behavior or automation.
PROJECT: DexOS — Identity-Stable AI Framework
Role: Founder / Lead Architect
Overview:
DexOS is a local-first AI framework designed to keep AI behavior, tone, and decision-making consistent across sessions, devices, and updates. The goal is to prevent the common issues seen in player-facing AI: personality drift, inconsistent advice, and unreliable behavior over time.
Key Contributions:
• Designed AI identity systems that preserve consistent behavior across resets and model changes.
• Built memory and context systems allowing AI to retain relevant information without hallucination.
• Implemented behavior controls that ensure reliability while remaining useful and responsive.
• Focused on long-session stability critical for live and service-based games.
Why this matters for games:
• Enables AI companions, assistants, or guides that feel consistent and trustworthy.
• Reduces player frustration caused by AI behaving unpredictably.
• Supports live updates without breaking AI behavior.
PROJECT: Guidance — Real-Time Gameplay Assist AI
Role: Architect / Systems Engineer
Overview:
Guidance is a real-time AI assistance system that analyzes live gameplay context and provides actionable insights to the player without automating gameplay or violating fair-play boundaries.
High-Level Functionality:
• Captures short snapshots of player-visible game state.
• Compresses context for fast AI reasoning.
• Returns clear, actionable guidance designed to support player decision-making.
Design Principles:
• Assist, don’t automate.
• Preserve player agency.
• Maintain consistent reasoning and tone.
• Avoid cheating or exploitative behavior.
Why this matters for games:
• Allows AI assistance without undermining skill or competition.
• Works for competitive, cooperative, and single-player games.
• Enhances player experience rather than replacing player control.
PROJECT: Edge & Local AI Deployment Stack
Role: Systems Engineer
Overview:
Designed AI systems capable of running locally or at the edge for low-latency, reliable performance in latency-sensitive environments.
Key Capabilities:
• Local inference pipelines optimized for responsiveness.
• Offline-capable AI behavior.
• Designed for predictable, stable outputs rather than
Hire a AI Engineer
We have the best ai engineer experts on Twine. Hire a ai engineer today.