DevOps Specialist at CIBC turned AI builder. I design autonomous agents, RAG systems and MCP servers on top of hardened data engineering — systems a team can actually depend on.
At CIBC I built a fully autonomous Jira-to-PR agent — it reads the ticket, writes the code, generates the tests, opens the PR, reviews itself, fixes its own errors, and moves the ticket through the dev → QA lifecycle. Work that took the team two full weeks now ships in under an hour.
Around it: a self-learning PR review system on ChromaDB that ingests feedback from every review and gets sharper with each one — raising code standards across a 9-person engineering team.
That runs on real engineering, not demos: MCP servers wired into Jira and Confluence, persistent agent memory with semantic retrieval, and a legacy-to-Delta Lake migration that made cluster processing 97% more efficient.
I work spec-first: define the behavior, write the test, then build. It's how I keep AI systems dependable instead of magical-until-they-break.
“Two weeks of development and testing — shipped in under an hour.”
— the Jira-to-PR agent, running in production at CIBCAn agent that ships its own pull requests
Case study →Answers grounded in private documents
Case study →Medallion architecture, tuned for speed
Case study →Open to AI Systems Architect and AI Engineer roles — remote or Greater Toronto Area.