yaya AI: AI-Powered End-to-End Job Hunt Automation platform

yaya AI automates your job hunt: it scans alerts, researches companies, and crafts ATS-optimized resumes and cover letters, delivering a personalized, ready-to-send Gmail draft in seconds.

Comment

💡 Inspiration

The job market in 2026 is more competitive than ever, and recruiters are often overwhelmed by generic, AI-generated applications. We wanted to build something that doesn't just "use AI" to write text, but acts as a true agentic partner. We were inspired by the idea of a "Digital Career Agent" that lives in your inbox, scouts for opportunities while you sleep, and uses professional HR domain expertise; not just LLM intuition to help you land your dream role.

🚀 What it does

yaya AI is an end-to-end autonomous job application system:

  • Auto-Scouting: It monitors your Gmail for job alerts and automatically populates a smart dashboard.
  • Deep Research: For every job, a dedicated Research Agent performs exhaustive company and culture lookups.
  • Professional Tailoring: It uses a specialized "Impact Formula" (Action Verb + Task + Metric) and ATS-optimization modules to rewrite your resume for every single application.
  • Full Automation: It converts materials into professional PDFs and creates a ready-to-send Gmail draft with all attachments included, personalized to the specific hiring team.

🛠️ How we built it

We leveraged the Google Agent Development Kit (ADK) to build a sophisticated multi-agent system:

  • Orchestration: A SequentialAgent (Job Search Lead) coordinates a team of specialized agents (Research, Tailor, Cover Letter, Converter, and Composer).
  • Brain: Powered by Gemini 3 Pro for complex reasoning and Gemini 3.0 Flash for fast, deterministic tool calling.
  • Infrastructure: Built with FastAPI and Firebase (Firestore/Storage) for state management.
  • Extensibility: We used MCP (Model Context Protocol) to create a "Studio Server" that gives our agents direct access to the filesystem and external APIs.

🚧 Challenges we ran into

  • Agent Recitation: Ensuring agents stayed focused on long-running tasks. We solved this by implementing a "Recitation TODO" system where agents update a scratchpad to maintain state awareness.
  • PDF Consistency: Balancing Markdown flexibility with strict resume formatting. We built a custom document conversion agent to handle the MD-to-PDF bridge.
  • Token Management: Handling large HR domain handbooks required careful context engineering to avoid losing the "signal" in the "noise."

🏆 Accomplishments that we're proud of

  • HR Logic Integration: Successfully embedding "Certified Professional Resume Writer" (CPRW) standards directly into the agent’s core instructions.
  • True Autonomy: The jump from "AI writing assistant" to an agent that actually does the work-scouting, researching, and drafting.

📖 What we learned

We learned that the power of Gemini 3 isn't just in the model's intelligence, but in its ability to follow complex, multi-step agentic workflows. We also gained experience in debugging agent-to-agent communication and the importance of structured logging for observability.

🔮 What's next for yaya AI

We plan to add interview preparation modules that use your tailored resume and the job description to generate simulated technical and behavioral interviews.

Built With

  • fastapi
  • firebase-(firestore-&-storage)
  • gcp/cloud-run
  • gemini-3-pro
  • gemini-3.0-flash
  • google(adk)
  • mcps
  • python
  • vertex-ai
Share this project:

Updates