At Automation Anywhere, I evaluated Agentic AI Digital Worker responses for hallucinations, accuracy, guardrail adherence, process deviation, and data privacy. I fed structured findings into improvement cycles that reduced process deviation from 75% to 7% within a year.
I also helped build the Automated QA system, translating audit logic, sampling rules, and scoring parameters into the automated pipeline. I built the organization's audit framework for Salesforce support cases and coached engineers on consistent evaluation standards.
My earlier work at Globsyn AI Workspace included managing data annotation and human-validation workflows for an Amazon AI/ML model-training program. In my own projects, I built an LLM-powered job-search agent and an on-device government form-filling assistant using MCP and A2A protocols.
