At Reddit, I designed bidding agents and refined deep neural networks for multi-conversion optimization in the advertising marketplace. I worked with noisy and delayed feedback and real-time calibration.
I developed online and offline evaluation metrics and investigated performance discrepancies, bias–variance tradeoffs, and the reliability of production deep learning models.
I analyzed full-funnel performance and marketplace dynamics to balance competing optimization objectives and inform modeling and bidding strategies. I also applied data quality and error-analysis practices to support reliable model outputs.
At Costockage, I built and maintained software systems focused on data handling and analysis using Python. I collaborated on data-driven projects that required careful validation and consistency checks.

