Building Sigma's AI copilot for workbooks (AI Builder) — an agent that uses natural language to create dashboards and applications. I helped design the core agent harness and helped build the evaluation infrastructure now used across all of Sigma's AI features.

I also led the App Logic team, building Actions — a low-code solution that transforms static BI dashboards into interactive workflows with background execution and AI integrations.

• Early engineer on the App Logic team, shipped Actions

• Built enterprise features: audit logging, authentication, OAuth integrations

Technical fellowship at SHV

• Held office hours and assisted students in debugging programs written in C using GDB and Valgrind

• Guided students in implementing low-level POSIX features such as a filesystem, shell, and malloc

• Maintained the CS 341 autograder used within the CS department to automatically grade assignments

• Deployed ML models to predict potential hardware and software failures in real time for over 23,000 internet devices that power the Azure network, aiming to increase overall reliability

• Incrementally trained XGBoost models on over 40GB of data daily by implementing batch training methods to prevent model drift and increase overall performance

• Generated new features by leveraging feature engineering and scheduled SQL functions to increase model performance to over 90% accuracy