Inspiration

Inspiration

I was recently laid off and I had to go to the local unemployment office to begin to receive benefits. As part of the identity verification process, they also make you create a resume and add it to the employment services site. So there I was, sitting in the unemployment office trying to verify my identity so I could collect unemployment benefits. The whole process was painfully dated - forms everywhere, confusing workflows, and I'm watching people struggle with basic resume tasks. I thought to myself, "The world deserves a better way than this."

That's when I decided to reimagine the whole process. With Google's free data limits, I realized I could offer this solution for free to people who really need it. I initially tried Google Docs but it wasn't a good experience - people needed something more intelligent, more responsive to their actual situations.

What it does

Get Back to Work With AI uses AI to turn your resume and job description into tailored resume to get through application tracking systems . The application listens in different languages and transcribes, making it accessible to a broader population who might struggle with traditional resume tools. It's designed to help unemployed individuals quickly create professional, tailored resumes that actually get results.

How we built it

This was a labor of love built across two laptops and phones, interacting with four to five AI agents on each device. I used everything I could get my hands on: GitHub Copilot, Jules, Loveable, Cline, Cursor, Windsurf, Aider, the free AI playground through Outlier.ai, and countless late nights with Claude. Codex was in the mix too, of course.

Get Back to Work With AI is powered by the Google Agent Development Kit, which orchestrates this series of AI agents (parsers, analyzers, enhancers, etc.). Deployment and function management are handled via Lovable's Git‑integrated tooling on Google Cloud. With 50% of my unemployment checks going to Anthropic API costs 😭 - but hey, that's the price of innovation when you're bootstrapping!

Challenges we ran into

Beyond the obvious challenge of building this while unemployed, deploying a multi‑agent system on Google Cloud proved incredibly challenging. Authentication, orchestration, and observability across dozens of serverless functions added complexity I hadn't anticipated. I overcame these hurdles by adopting Google's latest orchestration tools and refining our AI prompts to reduce errors and retries.

Accomplishments that we're proud of

  • Delivered a fully automated, end‑to‑end AI resume pipeline in MVP form
  • Achieved reliable, sub‑minute processing time per resume
  • Seamlessly integrated GADK, Lovable, and serverless functions for CI/CD
  • Enabled job‑specific tailoring with zero manual template editing
  • Built something that could actually help people in situations like mine at that unemployment office

What we learned

  • How to orchestrate multiple AI agents end‑to‑end with GADK
  • Rapid prototyping and continuous deployment using Lovable
  • Best practices for serverless architecture and observability on Google Cloud
  • Crafting effective prompts and retry logic for reliable AI workflows
  • That sometimes the best solutions come from your own pain points

What's next for Get Back to Work With AI

  • Add support for cover letters and portfolio websites
  • Build a real‑time feedback dashboard and analytics
  • Expand the library of resume designs and themes
  • Integrate collaborative editing and shareable links
  • Scale the free tier to help more people who are in the situation I was in

Built With

Share this project:

Updates