EzJobs

Inspiration

Job hunting is tedious, repetitive, and often emotionally draining—candidates spend countless hours tailoring resumes, researching companies, and tracking applications instead of focusing on interviews or skill growth. EzJobs automates these workflows via autonomous AI agents that collaborate to streamline every step of the career journey.

What it does

EzJobs is a cloud-hosted, AI-driven platform that orchestrates four specialized agents to handle the end-to-end application process:

Job Discovery Agent Scours multiple job boards with keyword, location, and remote filters Ranks opportunities by compensation, skill match, and employer reputation

Resume Tailoring Agent Parses each job description for key qualifications and keywords Generates customized resumes and cover letters that align with role requirements

Company Research Agent Aggregates insights on culture, Glassdoor ratings, and team structure Summarizes hiring manager profiles and provides commute or relocation data

How we built it

On Bolt Next!

frontend & Deployment

• React + TypeScript Via Bolt, deployed via Netlify
• Responsive, accessible UI with a dashboard

AI Agents & Messaging

• Each Agent is built with the Google ADK kit and deployed to GCP
• Utilizes the Supabase MCP server for operations
• Wrote custom tools in Python to generate a professionally formatted resume based on a template

Data & Storage

• Supabase PostgreSQL for structured data (job listings, user profiles, application logs)
• Supabase Storage holds all resume versions and user uploads in the resume-uploads bucket

Challenges we ran into

Having bolt around certainly did help reduce the burden of dealing with react, here are our challenges

• Resume Personalization: Crafting prompts that produced role-specific, human-like resume tweaks across diverse industries.

• Frontend Integration: Seamlessly embedding asynchronous agent results into a React UI without excessive polling.

• CORS headaches

• Bolt Github and project sync issues, we lost a lot of progress we made during the last week since the sessions didn't save and didn't commit to Github during this period, and even at submission time.

• Bolt doesn't know much about ADKs so we had to a lot of the leg work.

Accomplishments that we're proud of

That we got this to work in such a time crunch + busy lives

What we learned

• Prompt engineering is critical: small tweaks yield dramatically better AI outputs.
• How to develop and deploy AI Agents using an ADK framework and integrate them into a custom UI

• Don't be shy to try out tools such as Bolt for projects; when directed properly, it can improve productivity

  What's next for EzJobs

• Networking Agent: Auto-discover and outreach to relevant contacts, log responses, and schedule reminders.

• Browser Extension: Streamline applications directly from company websites using pre-filled profile data.

• Premium Insights: Add subscription tiers for deep-dive analytics, “insider” company trends, and live interview coaching.

Built With

• ai
• bolt
• google-adk
• google-cloud
• react