At Planview, I own the reliability roadmap for the enterprise AI assistant, deploying agentic workflows on customers’ own data. I built its evaluation framework and regression tests, raising correct-on-all-five-runs from 7 to 13 of 23 canonical customer questions.
I also analyze production traces to identify why the assistant fails, then turn those findings into product fixes and customer safeguards. One prompt-methodology change I developed shipped as a product feature; after testing showed it needed improvement, I rewrote its instruction and brought three real customer questions to five out of five.
Before that, I built Planview’s AI adoption practice, growing a pilot into a standard model across 13 cohorts and 8 enterprise customers. Earlier, I helped enterprise teams improve delivery flow and founded FCA’s Enterprise DevOps Platform Program. I co-authored Navigating the New Reality and write Uncommon Sensemaking.
