Software engineer building dependable products and AI systems.
I'm a Software Engineer 2 at Intuit, working on the QuickBooks Indirect Tax Platform in New York. I build across backend services, full-stack products, and production AI, with a focus on making complex systems understandable and reliable for the people who use them.
My day-to-day work spans Kotlin/Java services, Python AI tooling, React/TypeScript interfaces, and SQL.
- Products with demanding business logic. I've led tax engineering for QuickBooks' global invoicing rewrite and delivered reporting APIs and interfaces, connecting domain rules, financial correctness, and customer feedback.
- AI systems people can operate and trust. My production LLM work covers observability, governance, privacy controls, and cost visibility for an existing classification service.
- Infrastructure that connects transactions to decisions. I've built core transaction tax-data pipeline components using PostgreSQL CDC and Kafka to support reporting and filing.
- Tools that improve how engineers work. I build custom Claude Code skills and multi-agent workflows for research, root-cause investigation, and operational reporting.
I also taught and mentored students in CS 540: Introduction to Artificial Intelligence at the University of Wisconsin–Madison as a Peer Mentor, September 2023–May 2024.
These are earlier personal projects that show where I started: interactive algorithms, computer vision, and software that responds to the physical world. Most of my recent engineering work is professional and lives outside this public account; the repositories below do not contain Intuit code.
- Path Finder Visualizer — Explore breadth-first search on an editable maze, with a visual route from start to finish. Python / Tkinter.
- AutoDust — An early waste-sorting prototype connecting object detection, rule-based classification, and Raspberry Pi motor control. Python / OpenCV / YOLOv3. Historical hardware project; setup limitations are documented.
- Sudoku Solver — A small constraint-solving project with a reusable backtracking solver, command-line interface, and interactive board. Python / Tkinter.