AI Projects

DORA metrics are widely referenced in DevOps literature but rarely made accessible to Scrum Masters and PMs in a practical, team-facing format. The challenge was to build a tool that not only calculates the 4 key metrics but translates them into meaningful coaching guidance a Scrum team could act on immediately.

Designed and built a full-stack web application using FastAPI and Jinja2 for the backend, Chart.js for visual performance reporting, and the Gemini AI API for generating narrative summaries, team strengths, improvement areas, and retrospective coaching questions. Containerized with Docker and deployed to Google Cloud Run for production availability.

Delivered a fully functional, publicly accessible dashboard that classifies Scrum teams into Elite, High, Medium, or Low DORA tiers and generates AI-written sprint retrospective insights — demonstrating both technical delivery capability and deep Agile fluency in a single portfolio artifact.

As a Project Manager actively job searching across IT, Fintech, and major Canadian banks, I was spending hours every week manually searching LinkedIn, rewriting my resume for each posting, and drafting cover letters from scratch — repetitive work that didn't require my judgment, just my time.

Built an ADK 2.0 graph workflow that searches LinkedIn via SerpAPI, then pauses for my review before generating anything — I choose which jobs are worth pursuing, and only then does Gemini 2.5 Flash tailor a resume and cover letter in parallel for each selected role. Documents are emailed directly to my inbox. A FastAPI dashboard on Cloud Run serves as the control interface.

Delivered a fully functioning agent system — deployed to Vertex AI Agent Runtime with Terraform-provisioned infrastructure — that finds fresh LinkedIn postings daily, generates ATS-optimized application documents, and delivers them by email. Built end-to-end as my first AI engineering project, using Google Antigravity as my development environment.