PipelineIQ

• 

  PipelineIQ — Gigi the Chihuahua AI agent diagnosing your GitLab pipelines

• 

  Issue report

InspirationPipelineIQ 🐾

Inspiration

Every developer has been there — it's 2am, your pipeline just failed, and you're drowning in hundreds of lines of cryptic logs. I wanted to build something that could take that pain away automatically. That's how Gigi was born: a sassy Chihuahua AI agent who never sleeps and always has your back.

What I Learned

• How to build a real agentic workflow using Google Gemini 2.5 Flash
• Integrating GitLab's API to read pipeline logs and create issues programmatically
• Deploying a production FastAPI + ASGI service on Render
• The importance of separating local proxy settings from production environment config (learned the hard way 😅)

How I Built It

PipelineIQ is built with FastAPI as the backend, Google Gemini 2.5 Flash as the AI brain, and python-gitlab for GitLab integration. When a webhook fires, Gigi fetches the failed job logs, sends them to Gemini for root cause analysis, and automatically creates a GitLab Issue with a diagnosis report and a hotfix branch.

Challenges

The biggest challenge was network configuration — getting Gemini API calls to work correctly in a cloud production environment after developing locally behind a proxy. Debugging ASGI vs WSGI compatibility with Gunicorn was another fun surprise at midnight.

Built With

• 2.5
• api
• fastapi
• flash
• gemini
• gitlab
• google
• gunicorn
• python
• render
• uvicorn