DEV Community

Cover image for A Pragmatic AI Portfolio Built with Gemini & Cloud Run
Luke Ponga
Luke Ponga

Posted on • Edited on

A Pragmatic AI Portfolio Built with Gemini & Cloud Run

#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

I’m Luke Ponga, a software developer and IT support specialist based in Hamilton, New Zealand.

My background in IT support and systems troubleshooting strongly influences how I approach AI. I build for real users, real constraints, and real workflows—not demos or novelty experiences.

Portfolio

🔗 Live Deployment:

https://portfolio-luke-706599404958.us-central1.run.app/

AI at the Core (What Makes This Portfolio Different)

This portfolio treats AI as a first‑class system, not an add‑on.

Instead of embedding a generic chatbot, I used Google Genkit and Gemini models to build task‑specific AI flows that are:

  • context‑aware
  • grounded in real project data
  • constrained to professional, explainable outputs

AI here acts as a guided assistant, not a black box.

Key AI Features

🤖 AI Project Matcher (Interactive Portfolio Guide)

This feature allows visitors to explore my work using natural‑language queries, such as:

  • “Show me AI projects”
  • “What experience do you have with Gemini or Genkit?”
  • “Which projects are most relevant for backend roles?”

How it works:

  • Implemented as a Genkit flow with structured context injection
  • Uses curated project metadata rather than free‑form generation
  • Responds only within known scope, reducing hallucinations
  • Optimised for fast, low‑latency responses using Gemini 1.5 Flash

🧠 Career Architect (AI‑Assisted CV & Cover Letter Generator)

  • Accepts a user‑provided job description
  • Generates tailored CVs and cover letters in Markdown
  • Enforces professional tone and structure
  • Powered by Gemini 2.0 Flash via Genkit

How I Built It

Architecture & Performance

  • Next.js 16 (server‑first rendering)
  • Containerised and deployed on Google Cloud Run
  • Optimised for performance and SEO

AI Engineering

  • Google Genkit for deterministic AI flows
  • Strong separation of context, prompts, and formatting
  • Middleware to reduce hallucinations and enforce tone

What I’m Most Proud Of

This portfolio treats AI as a responsible engineering tool, not a novelty.
It remains usable even without AI enabled.

Technical Details

Thanks to DEV and Google AI for encouraging real‑world, responsible AI use.

devnewyear2026 #googleai #cloudrun #nextjs

Top comments (0)