I Finally Updated My 2-Year-Old Portfolio (Thanks to Antigravity)

#devchallenge #googleaichallenge #portfolio #gemini

New Year, New You Portfolio Challenge Submission

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

About Me

Hey! I'm Arjun. I'm 16, from Lucknow, India. I build stuff on the web, write about tech, and I'm the CTO at Pilot.

I've shipped 20+ projects, have 34,000+ followers on DEV.to, and made $1,375 from technical writing. My research project got selected for INSEF Regional Fair too. This portfolio is basically proof that I actually do things.

Portfolio

How I Built It

The Problem

My portfolio was 2 years old. That's ancient in tech time. The projects were outdated. The design was ugly. It needed to go. Here's the old one if you want to see what I was working with.

The last commit before this rebuild? September 15, 2024. Then nothing until January 2, 2026. By January 7, the whole thing was done. 5 days. Here's the PR if you want to see the chaos:

66 commits
78 files changed
+15,591 lines added, -9,667 lines removed

Yeah. It was a lot.

The Fix

I used Antigravity, Google's AI IDE. It made the whole thing way easier than I expected.

Here's what I did:

Added new projects and achievements
Removed old stuff that didn't represent me anymore
Made the design look actually good with animations and dark mode
Cleaned up the clutter

The funny part? I did most of the data updates through Antigravity too. Like, my resume.tsx file that runs the whole site? I just told the AI what to change and it changed it. Same with the markdown files for my chatbot.

I barely typed anything. I just described what I wanted. True vibe coding. At this point, I'm not even sure if I can call myself a developer or just a really good prompter.

Tech Stack

Next.js 16 with App Router
TypeScript
Tailwind CSS and shadcn/ui
Framer Motion for animations
Google gemini-3-flash for the AI stuff
Vercel Postgres with pgvector for vector search
Deployed on Google Cloud Run

What I'm Most Proud Of

The RAG Chatbot

So after I finished updating everything, I asked Antigravity what else I should add.

It said: "RAG Chatbot."

I said: "Okay, let's do it."

Then Antigravity built it. And it actually works. Go try it on the portfolio.

I also used AI Studio to test prompts. I wanted the chatbot to sound like me, not some generic AI. That's also where I got my API key from.

You can ask the chatbot anything about me. My projects, my background, my skills. It gives real answers based on actual facts about me. No made up stuff.

Here's how it works:

I have markdown files with info about me in src/data/rag/
These get turned into vector embeddings using Google Gemini's text-embedding-004 model
When you ask a question, your question also becomes an embedding
The system finds the most similar chunks from my knowledge base using Postgres and pgvector
Those chunks plus your question go into gemini-3-flash
The model writes a response based on the real facts and streams it back