At Draup, I took an IT spend estimation engine from 0 to 1, turning a manually produced, request-based service into a daily pipeline across 36K+ companies. It’s now used by 10–20 enterprise accounts, and I shipped a phase-two spend-by-category breakdown.
I also owned the financial data product end-to-end, from ingestion to customer-facing output, across 10+ financial attributes. That data powers company intelligence, benchmarking, alerts and enterprise sales workflows.
For Pulse, Draup’s AI intelligence agent, I framed the problem and owned the vision, workflow and PRD. I prototyped flows, coordinated the build through Jira epics and acceptance criteria, and evaluated the agent on real customer configurations before release.
At Draup, I also shipped a deterministic annual-report matching approach that reached 92% coverage across 17.5K reports without new data spend. Earlier, at Raspa Pharma, I analysed 4M+ reactor data points into decision dashboards and built a metabolic model with 90% accuracy; independently, I built an AI nutrition tracking app with a free-text logging flow.
