🚀 My 2026 AI-Powered Portfolio: Built with Google Antigravity & Deployed on Cloud Run

Mangesh Raut — Mon, 19 Jan 2026 07:38:52 +0000

This is a submission for the New Year, New You Portfolio Challenge Presented by Google AI

About Me

Hi, I'm Mangesh Raut, a Software Engineer and aspiring AI/ML expert currently pursuing my Master's at Drexel University.

I built this portfolio to be more than just a resume—I wanted it to be an intelligent, interactive experience that reflects who I am as a developer in 2026. My goal was to create a space where visitors could naturally converse with an AI representation of my professional persona, explore my work through dynamic visualizations, and even have a bit of fun with a hidden game.

By integrating Google's latest AI tools with a modern, glassmorphism-inspired design, I hope to demonstrate that personal portfolios can be powerful, agentic applications that truly "live" on the web.

Portfolio

Alternate Link: https://mangesh-portfolio-api-239037094122.us-central1.run.app

Note: The deployment is hosted on **Google Cloud Run* and features a custom dev-tutorial label.*

How I Built It

To build this next-generation portfolio, I leveraged the full power of the Google AI ecosystem within a modern full-stack architecture.

🛠️ Tech Stack

Frontend: HTML5, Tailwind CSS 4.x (Glassmorphism UI), JavaScript ES2024+
Backend: Python 3.12+, FastAPI (high-performance ASGI)
AI Engine: Google Gemini 2.0 Flash (via AI Studio) & Grok (Assistant Context)
Dev Environment: Google Antigravity (AI-First IDE)
Deployment: Google Cloud Run (Serverless Containers)

⚡ Powered by Google AI Tools

Google Antigravity: This was my "co-pilot" for the entire project. It helped me:
- Prototype 10x faster: Generating the initial project structure and boilerplate code.
- Debug instantly: Fixing complex async/await issues in the streaming chat logic.
- Design with code: creating the CSS for the elaborate glassmorphism effects and animations.
- Refactor: Keeping the codebase clean and modular as features grew.
Google Gemini API (via AI Studio):
- I integrated Gemini 2.0 Flash to power the "AssistMe" chatbot. Its speed and low latency make it perfect for real-time, streaming conversations.
- The API handles multi-turn context, allowing the AI to remember previous questions (e.g., "What are his skills?" -> "Which ones is he best at?").
- I used system prompting to give the AI a specific persona: a helpful, professional guide to my career.
Google Cloud Run:
- Deploying the containerized Python app was seamless. The auto-scaling ensures the site handles traffic spikes (like from this contest!) without breaking.
- I added the required label: gcloud run deploy ... --labels dev-tutorial=devnewyear2026.

🧩 Key Design Decisions

Glassmorphism: I chose a frosted-glass aesthetic (Apple-inspired) to feel premium and futuristic.
Agentic UX: Instead of just distinct pages, the AI assistant acts as a comprehensive navigation tool, capable of "taking actions" like toggling themes or scrolling to specific sections.
Performance First: Despite the heavy visuals, the site hits 90+ Lighthouse scores thanks to optimized assets and fast server-side responses from Cloud Run.

What I'm Most Proud Of

There are three key features I'm incredibly proud of shipping in this challenge:

1. 🧠 "AssistMe" — The Neural AI Assistant

This isn't just a generic chatbot. It features:

Character-by-character streaming: It feels like it's thinking and typing live.
Agentic Capabilities: It can actually control the website. Ask it to "Show me his projects" or "Turn on dark mode," and it will execute the action via code.
Context Awareness: It knows exactly who I am and answers questions based on my resume data.

2. 🎮 Debug Runner Game

I built a fully functional HTML5 Canvas mini-game hidden in the "Game" section. It features:

Smooth 60fps animations.
Mobile touch controls.
Retro-inspired graphics. It was a fun technical challenge to implement a game engine from scratch alongside a React-like component system!

3. 🎨 The "Live" GitHub Section

Instead of static project cards, I integrated the GitHub API to pull my latest repositories in real-time. The portfolio automatically filters out forks and sorts by the most recently updated, so my "Latest Projects" section is always up to date without me touching the code.

Resources:

Live Cloud Run Demo: mangesh-portfolio-api-239037094122.us-central1.run.app
GitHub Repository: github.com/mangeshraut712/mangeshrautarchive

Built with ❤️ in Philadelphia using Google Antigravity & Cloud Run.

DEV Community: Mangesh Raut