At Oracle, I built and owned CI Insights, a Python agent on OCI Generative AI that classifies CI failures and proposes fixes. Human reviewers accepted about 60% of its fixes, and it cut failure-to-fix time by about 40%.
As Fusion CRM moved to agentic AI, I tested the Account Advisor and Deal Advisor agents for answer accuracy, reasoning, security, and the actions they take. I also defined risk-based release gates across 10+ Scrum teams, raising on-time releases from about 70% to over 90%.
On Layam, I created a 38-case golden dataset and wired LLM evaluations into CI to block merges below score thresholds. I’ve also built RAG pipelines in projects and training sessions, and bring experience testing enterprise applications in banking, insurance, and utilities.

