This is a submission for the New Year, New You Portfolio Challenge Presented by Google AI
About Me
I am a Principal Product Architect and Google Developer Expert (Cloud) with over 12 years of experience in designing scalable, security-first cloud ecosystems. My goal with this portfolio was to move beyond a static résumé and create a "System Architect’s Console"—a live, gamified environment that reflects my expertise in migrating complex monoliths to containerized, AI-driven architectures. I wanted to show how a senior-level perspective can leverage Google AI to solve enterprise-scale problems like compliance and developer productivity.
Portfolio
How I Built It
My development process leveraged a minimal dependency surface area and advanced AI-driven workflows:
1. The "AI-First" Development Process
This project was developed in partnership with Antigravity, an agentic AI assistant:
- Iterative Design: I used Antigravity for the dynamic creation and refinement of the "Obsidian" CSS theme and responsive layouts.
- Automated Verification: I utilized Antigravity's browser sub-agents to perform real-time testing of custom logic, such as popover positioning and chat accuracy.
- Architectural Guardrails: Antigravity assisted in implementing the Firestore caching layer and serverless deployment pipelines.
2. The Tech Stack
- Backend: Zero-Framework Python using the standard http.server library to minimize container size (under 150MB) and reduce cold start latency.
- Frontend: Premium Obsidian-Style UI built with Vanilla HTML/CSS/JS (no heavy frameworks).
- Database: Firestore (Datastore Mode) using a Cache-Aside Pattern with raw REST API calls.
- Cloud Infrastructure: Hosted on Google Cloud Run
- CI/CD: Deployed via Google Cloud Build, pushing images to Artifact Registry.
3. Google AI Integration
AI is at the core of the user experience:
- Vertex AI (Gemini 3.0 Flash Preview): Powers the "AI Chat Consultant" and profile summarizer.
- Grounding & Data Fencing: The AI is grounded using a PDF resume on Google Cloud Storage (GCS) and uses "Hard Boundary" prompts to prevent information cross-contamination.
- Prompting Strategies: Integrated "Strict Data Fencing" and an "Agentic Professionalism" executive persona for the AI consultant to ensure PII security.
What I'm Most Proud Of
I am most proud of the Standard Library Power and Cost Efficiency achieved in this build.
By avoiding heavy frameworks and using Firestore to cache AI responses at the edge, I reduced potential API costs by over 80% for common navigation paths. Furthermore, the Intelligent Positioning logic in Vanilla JS—which dynamically calculates viewport boundaries to prevent content clipping—proves that professional, high-fidelity UX can be achieved through pure engineering without the overhead of modern frontend frameworks.
Top comments (0)