Overview
My AI-focused work emphasizes architectural understanding over tool fluency — exploring how generative AI, agent-based systems, and intelligent automation alter system design, governance, and long-term sustainability.
Rather than building consumer-facing demos, I focus on:
- Reference implementations that demonstrate architectural patterns
- Security and trust boundaries in AI-enabled systems
- Integration of AI components into enterprise architectures
- Practical application of formal AI education to real-world systems
- Model, Tool, and MCP Server selection, implementation and Usage strategies
Active Projects
AI Portfolio — Certification Labs & Research
Repository: steven-tomlinson/ai-portfolio
Status: Active Development
A structured portfolio containing:
- Certification Lab Projects — Hands-on implementations from Google Cloud and Microsoft Azure AI certifications
- Courseware & Artifacts — Study materials, course notes, and lab exercises
- Personal Experiments — Testing AI concepts, libraries, and integration patterns
- Learning Notes — Observations on architectural implications of AI systems
Technologies:
- Azure AI Services and Azure OpenAI Service
- Azure Bot Service and Azure Functions
- HTML/CSS/JavaScript for portfolio presentation
- VS Code with GitHub Copilot
Focus Areas:
- AI agent architecture and orchestration
- Azure AI Foundry certification path
- Safe, secure implementation patterns
- Creating novel abstractions for complex system design
Lockb0x Codex Forge — Digital Provenance System
Repository: steven-tomlinson/lockbox-codex-forge
Status: Production-Ready (Awaiting Marketplace Publication)
A Chrome Extension (Manifest V3) implementing the lockb0x protocol for creating secure, verifiable Codex Entries from web content or user-uploaded files.
Key Features:
- Complete lockb0x protocol implementation (UUID generation, SHA-256 hashing, ni-URI encoding, JSON canonicalization, ES256 signing)
- Zip archive workflow with dual-signature verification
- Google Drive integration for secure anchor storage
- Support for any file type (text, PDF, JSON, binary)
- Schema validation against lockb0x schema v0.0.2
- Encrypted zip archiving with verifiable provenance
Technologies:
- JavaScript (Manifest V3 Chrome Extension)
- Google Drive API and OAuth2 authentication
- Cryptographic primitives (SHA-256, ES256)
- JSON Web Keys (JWK) and JSON canonicalization (RFC 8785)
Architectural Focus:
- Trust boundaries and verifiable digital provenance
- Protocol-driven system design
- Secure credential handling and token management
- User-controlled data sovereignty
Note: AI-powered metadata generation features are referenced but deprioritized, as Chrome Built-In AI APIs are still experimental. Current implementation uses fallback text extraction.
Roadmap:
- Final release to Google Chrome Marketplace
- Fork for Microsoft Edge and OneDrive integration
- Enhanced testing infrastructure and integration tests
Liquidity Dashboard — DeFi Analytics (Early Stage)
Repository: steven-tomlinson/liquidity-dashboard-
Status: Template/Early Development
A Python-based Streamlit application for decentralized finance analytics and liquidity monitoring.
Technologies:
- Python and Streamlit
- Data visualization and analytics
Context: This project applies learnings from the Duke University DeFi specialization to practical financial data analysis and monitoring.
Educational Foundation
All project work is grounded in formal education:
Google Cloud — Generative AI Leader Specialization (2025)
Completed specialization covering:
- Organizational adoption of generative AI
- Agent-based system architecture
- Enterprise integration patterns
- Governance, risk, and value creation
- AI application development and transformation
Architectural Principles
My AI and protocol work follows consistent principles:
- Security First — Never commit secrets, design with trust boundaries explicit
- Verifiable Systems — Cryptographic proofs over implicit trust
- User Sovereignty — Users control their data and credentials
- Enterprise Integration — Systems must integrate cleanly with existing enterprise architectures
- Long-term Sustainability — Solutions designed to evolve without rewrites
Tools & Methodologies
Development Environment
- VS Code with GitHub Copilot
- Azure services and Google Cloud Platform
- Chrome DevTools for extension development
- Git and GitHub for version control
Architectural Approach
- Reference architectures over one-off solutions
- Explicit decision boundaries and failure modes
- Protocol-driven design where appropriate
- Test-driven development for critical paths
Future Directions
Ongoing exploration includes:
- Multi-agent orchestration — How autonomous agents coordinate and maintain trust
- Hybrid architecture patterns — Integrating decentralized components into enterprise systems
- AI governance frameworks — Policy-driven systems for intelligent automation
- Protocol evolution — How protocol-backed systems maintain backward compatibility
Code Quality & Best Practices
All projects emphasize:
- Clean, well-documented code
- Comprehensive testing where infrastructure exists
- Security-first design (no credentials in source control)
- Regular commits with meaningful messages
- Consistent use of templates and organizational patterns