Where I'veshipped.

Backend lead, Luminara

Led the backend of an app that makes academic reading accessible: dense text in, clear text out, with the meaning kept.

• Built most of the Flask backend: file conversion for PDFs, slides, Word docs and photos (with OCR), and back to the original format.
• Designed the pipeline so a local AI model rewrites only flagged spans, never the whole document, after NLP finds the hard parts.
• Built secure sign-in with JWT, disposable-email blocking and login rate limiting, plus a personal-dictionary API.
• Helped teammates with their code and Figma designs, and shipped it all with Docker Compose.

Asset tracking for field teams

About 800 cameras, teams across the country, and no single place to see where anything was. So I built one.

• Built a centralized web-based asset management system for about 800 GoPro cameras used by field teams across the United States.
• Designed an asset identification and allocation workflow that tracks who owns each camera, whether it's available, and which job site it's on.
• Worked from the company's San Francisco office to learn the real workflow, then iterated on the system from real-world use.

Features for 30,000+ students

Shipped a feature for a platform the whole engineering student body can use, and a tool that cut most of the repeated config work.

• Built an end-to-end professor review feature for a platform serving 30,000+ students, combining student reviews with Rate My Professor data from a GraphQL/Python scraping pipeline.
• Led a 3-person team to build a Python/Grafanalib wrapper for server monitoring: 10 dashboards with about 70% less repeated time-series config code.
• Switched from Python into React/JavaScript to ship inside a 12-person engineering team, helping cut duplicate work across features.

Teaching models to see

My first real AI work: cleaning large image datasets and training models to classify them, before I'd even started college.

• Built end-to-end Python pipelines with Pandas and OpenCV to process and clean large image datasets.
• Implemented and tuned image-classification models, reaching over 85% validation accuracy.
• Automated data-visualization dashboards summarizing key image metrics, so project leads got insights faster.

Founded a studio for local shops

A San José studio that gets small businesses found online: websites, Google review stands, Google profiles and Instagram.

• Built the brand, domain, email and a hand-coded website with a custom physics engine that holds 60 fps.
• Designed and wrote a 9-post Instagram launch and a tap-or-scan Google review stand product.

Showing students that good design is for everyone

• Hosted 5+ UI/UX workshops for SJSU students using Adobe Express.

Leadership accelerator

• Completed a semester-long leadership accelerator (SSCI 101), coached by professionals in a cohort of 10 peers.
• Built career skills through mock interviews, networking and a capstone project with employer feedback.

Member

A campus community for students building new ideas and ventures.